LL1-04 hindsight check: Benzene (Infante), Late lessons from early warnings (EEA, 2001), Ch. 4, pp. 38–51#

Checked 25 September 2026. The check covers developments from 2001 to September 2026, with a few pre-2001 items the chapter did not cite, where they help weigh a claim. Page numbers refer to the 2001 report.

Overview#

The chapter makes two kinds of argument. The first is historical: benzene’s toxicity was known for decades while exposure limits stayed above the evidence, and US regulation was delayed and then constrained. Nothing published since 2001 challenges this. The second is a set of live claims from 2001. These covered the adequacy of the 1 ppm limit, the range of diseases benzene causes, risk at low doses, the tactics of industry consultants, benzene in gasoline and consumer products, and the relevance of precaution. The record since 2001 splits these into three groups.

Vindicated in direction, often more strongly than Infante could show in 2001. - Almost every expert and regulatory body outside US federal regulation has moved well below 1 ppm: - The EU binding limit fell to 0.2 ppm from 5 April 2026. - The EU’s scientific committee (ECHA RAC) recommended 0.05 ppm in 2018. - ACGIH cut its recommended limit from 0.5 ppm to 0.02 ppm in the 2024 cycle. - NIOSH’s recommended limit remains 0.1 ppm. - An industry consortium proposed 0.25 ppm in 2020. - OSHA’s permissible exposure limit is still the 1987 value of 1 ppm. OSHA itself says many of its limits are “outdated and inadequate”. - In 2025 IARC classified automotive gasoline itself as carcinogenic to humans (Group 1), causing acute myeloid leukaemia and bladder cancer. - The 1978 “voluntary” withdrawal from consumer products turns out to have followed a proposed federal ban. That ban was dropped in 1981 on the basis of information from industry.

Mixed or weakened where Infante went furthest. - Range of diseases. Later reviews firmly accept benzene as a cause of acute myeloid leukaemia (AML). Most also accept an association with myelodysplastic syndrome (MDS). IARC (2017) rates non-Hodgkin lymphoma, chronic lymphocytic leukaemia, multiple myeloma and chronic myeloid leukaemia only as “positive associations” (limited evidence). It dropped adult acute lymphoblastic leukaemia from that list. - The group’s own reanalysis. The NCI–China group, whose studies Infante relied on, rebuilt its exposure estimates. It then found little exposure–response for non-Hodgkin lymphoma and other lymphoid cancers. - Size of the risk. Infante’s estimate of 54 excess deaths per 1,000 workers at 1 ppm is roughly 5–50 times higher than agency estimates, which cluster around 1–10 per 1,000. - Deaths from the delay. The 198 + 77 deaths attributed to the 1977–87 delay remain an unvalidated projection. The myeloma part now rests on weaker evidence.

Confirmed as a pattern. Industry-linked contestation continued as Infante predicted: - a Chinese research programme funded by members of the American Petroleum Institute (API), 2001–2009; - consultancy meta-analyses finding no link with lymphoma or chronic myeloid leukaemia; - a critique of the 2004 low-dose blood-toxicity study; - in 2026, API-funded critiques of IARC’s gasoline classification.

But industry-linked work also produced evidence of risk at low doses, and an industry proposal for a limit below 1 ppm. The picture is contested, but it is also converging.

Weight for the lens. The chapter’s strongest lessons are about institutions, not toxicology: - knowing is not acting; - limits set by feasibility trail the evidence; - the burden of proof is placed on the regulator; - diffuse exposure outlasts workplace controls.

The later record supports all four, and in the US case strongly. The chapter’s quantitative and disease-spectrum claims should be cited with care, as one protagonist’s upper-end reading. The author was the lead investigator on the first benzene cohort study and continued writing on benzene after 2001.

2013 update. Late lessons II (EEA, 2013) did carry a benzene update. It is a two-page note by Infante in Annex 3, “Benzene and gasoline: an update on risk of adult and childhood cancers and issues with quantitative risk assessment” (2013 report, pp. 720–721). I read it from an archived copy of the EEA’s annexes file (see Sources). It: - repeats the 2001 list of cancers (“all of the major forms of leukemia”, plus NHL, myeloma and CLL); - cites newer studies showing raised AML and myeloma risk at average exposures below 1 ppm, and blood effects down to 0.1–0.2 ppm; - argues that gasoline “should be considered a cause of all lymphohematopoietic cancers associated with benzene exposure to both adults and children”; - adds childhood leukaemia from exposure in the womb, from home painting and from living near petrol stations; - argues that risk assessments underestimate risk at environmental levels. It gives two reasons: low-dose metabolism, which may make them 3–9 times too low, and leukaemia deaths missing from the Pliofilm cohort (4–9 extra, per two papers then “submitted for publication”).

Annex 2 of the 2013 report also re-presents the 2001 case. Table A2.4 (p. 705) repeats unchanged the line “This 10-year delay caused more than 200 deaths”. Table A2.1 (p. 702) dates benzene’s “effective risk reduction action” to the 1978 voluntary withdrawal, giving 81 years of inaction. The 2013 update therefore doubles down on the 2001 positions. It does not revise them. How those added claims fared is noted under Claims 2, 3, 4, 8, 9 and 10 below.

Claim 1. The 1 ppm OSHA limit is inadequate on cancer risk alone (p. 41)#

Original claim. - Infante’s 1997 risk assessment used the NCI/CAPM (National Cancer Institute–Chinese Academy of Preventive Medicine) data. It “suggests an extra risk of 54 deaths from leukaemia/lymphoma per 1 000 workers” over 45 years at 1 ppm. That is “54 times greater than a level considered significant by OSHA” (p. 41). - The chapter says those data “clearly demonstrate the inadequacy of the 1 ppm exposure limit… based on cancer risk alone” (p. 41). - It also records that OSHA’s own 1987 assessment left 10 extra leukaemia deaths per 1,000 at 1 ppm, and that the limit was set on economic feasibility (p. 41). - The 1997 source was a proceedings paper for the Toxicology Forum, not a peer-reviewed article (reference list, p. 49).

What happened since.

Exposure limits: where they stand now.

TWA is the 8-hour average limit; STEL is the short-term (15-minute) limit.

Further points on the limits: - OSHA’s own view. OSHA states that “many of its permissible exposure limits (PELs) are outdated and inadequate for ensuring protection of worker health.” - Exempt US sectors. Industries exempt from the benzene standard still fall back on the older limits: 10 ppm TWA, a 25 ppm ceiling and a 50 ppm peak (Table Z-2). - Possible further EU cut. In recital 26 of the 2022 directive, the Commission undertakes to assess the feasibility of going lower, “taking into account the RAC opinion of 2018”. The tripartite advisory committee (ACSH) advised starting that assessment by January 2028. - How the EU chose 0.2 ppm. The binding value came out of a negotiated opinion of the ACSH, whose members represent governments, employers and workers (4 June 2019). - Employers asked for three extra years for foundries. - Workers asked for RAC’s 0.05 ppm by 2030. - Industry’s own proposal. Authors publishing on behalf of the Lower Olefins and Aromatics REACH Consortium (an industry consortium) derived an occupational limit of 0.25 ppm in 2020 (Schnatter et al., 2020).

Risk estimates at 1 ppm. No later agency adopted a figure close to 54 per 1,000:

• RAC’s approach. RAC did not use linear low-dose extrapolation at all. It judged a mechanism-based threshold (for chromosome damage) defensible. It called linear cancer-risk estimates “overly conservative”, while noting “a remarkable consistency of published cancer risk estimates” at exposures above 1 ppm.
• Consensus on inadequacy. Most estimates, other than Japan’s (which sits at the benchmark), exceed OSHA’s own significance benchmark of 1 per 1,000. Every body that has revised its value since 2001 went well below 1 ppm.

Exposure in practice. Actual exposures in most high-income workplaces are already below 1 ppm: - IARC Vol. 120 reports that full-shift exposures “are usually less than 1 ppm”. - RAC puts typical EU workplace averages below 0.1 ppm. - Higher exposures occur in maintenance, tank cleaning and similar tasks.

This bears on Infante’s own remark that most US settings reach 0.2–0.3 ppm (p. 41). A limit of 1 ppm mainly fails to protect people doing high-exposure tasks and people outside the standard’s scope.

Verdict: partly held up. - Strengthened: the qualitative conclusion that 1 ppm is not protective against cancer, judged by OSHA’s own benchmark. It is now the international expert and regulatory consensus outside the US federal rule. - Not corroborated: the headline figure of 54 per 1,000. It sits well above every agency estimate found. - Unchanged: the US limit itself.

Weight. Cite “1 ppm is inadequate” as well supported and now mainstream. Do not cite “54 per 1,000” or “54 times” as established. It is one author’s upper-end estimate, and the chapter does not flag it as such. The fact that the US limit has stood unchanged for 39 years is itself strong evidence for the chapter’s lesson on institutional inertia (p. 46).

Claim 2. Benzene causes a broad range of blood and lymphatic cancers (p. 42; pp. 44–45)#

Original claim. Benzene is linked to: - “all major forms of leukaemia”: acute myeloid (AML), acute lymphoblastic (ALL), chronic lymphocytic (CLL) and chronic myeloid (CML); - hairy cell leukaemia, MDS and myeloproliferative disorders; - non-Hodgkin lymphoma (NHL), “including multiple myeloma” (p. 42).

Consultants who argued that benzene causes only AML are presented as out of step with the literature (pp. 44–45).

What happened since.

• IARC Vol. 100F (meeting 2009, published 2012). “Benzene causes acute myeloid leukaemia/acute non-lymphocytic leukaemia.” It notes “a positive association” with ALL, CLL, multiple myeloma and NHL.
• Most cohort studies showed no link with multiple myeloma.
• The earlier Pliofilm excess “did not persist in the most recent update”.
• The NCI–China cohort was a “particular exception” in showing NHL.
• IARC Vol. 120 (meeting October 2017, published 2018). Benzene “causes acute myeloid leukaemia in adults”. Positive associations are listed for NHL, CLL, multiple myeloma, CML, AML in children and lung cancer.
• A small minority of the Working Group judged NHL to be caused by benzene.
• For adult ALL, “data… remain sparse”, and all confidence intervals included no effect. It is no longer on the positive-association list.
• The evaluation does not mention hairy cell leukaemia, MDS or myeloproliferative disorders.
• MDS.
• The NCI group (Linet et al., 2019) refers to “more recent evidence of association with myelodysplastic syndromes”.
• Industry-affiliated pooled petroleum studies found MDS risk at low cumulative exposure (Schnatter et al., 2012).
• RAC (2018) accepts “evidence of an association” with MDS.
• Pliofilm cohort update (Rinsky et al., 2002). Of the new myeloma deaths, “three of [four]… were in workers judged to be unexposed”. Leukaemia risk was reaffirmed.
• NCI–China cohort, 28-year follow-up (Linet et al., 2015). Significant excesses:
• MDS/AML: risk ratio (RR) 2.7;
• NHL: RR 3.9;
• all lymphoid leukaemia: RR 5.4;
• lung cancer: RR 1.5.

Chronic myeloid leukaemia was also raised (RR 2.5) but not significantly.

The authors describe “increased risks of a broad range of myeloid and lymphoid neoplasms”. This supports Infante’s breadth, but only in comparisons of exposed with unexposed workers. - The same group’s exposure–response analyses, using rebuilt exposure estimates (Portengen et al., 2016): - MDS/AML: a suggestive trend (p = 0.08), strongest for exposure within the previous 10 years and for workers first exposed before age 30 (Linet et al., 2019). - Lymphoid cancers: “little evidence of exposure-response for benzene and NHL, [lymphoid leukaemia], ALL, or total [lymphoid neoplasms]” (Linet et al., 2020). - Other studies on lymphoma. - A women’s cohort in Shanghai found an NHL exposure–response (Bassig et al., 2015). - A meta-analysis funded by the US National Institute of Environmental Health Sciences (NIEHS) supports a causal link with NHL, especially diffuse large B-cell lymphoma (Rana et al., 2021). - Meta-analyses by consultancies reach the opposite conclusion for NHL and for CML (Alexander et al., 2010; Weed, 2010; Lamm et al., 2005 and 2009). - Low-exposure petroleum studies. No convincing association was found for CML or myeloproliferative disease (Glass et al., 2014). Hairy cell leukaemia showed a non-significant raised risk in the 2021 meta-analysis (Rana et al., 2021). - The 2013 EEA update. Annex 3 of Late lessons II (p. 720) restated the full list: “all of the major forms of leukemia, such as [AML, ALL, CML], non-Hodgkin’s Lymphoma (NHL), multiple myeloma (MM) and chronic lymphatic leukemia (CLL)”. It cited mainly Infante’s own work and a handbook chapter he co-wrote (Infante and Bingham, 2012). This came four years after IARC’s 2009 working group had declined to go that far. Infante had published a critique arguing that the 2009 evaluation “was incomplete and needs to be reconsidered” (Infante, 2011). IARC’s 2017 re-evaluation still did not accept the broader list (Vol. 120, above).

Verdict: partly held up. - AML is settled. MDS is widely accepted. - NHL, CLL, multiple myeloma and CML have moved from assertion to “limited evidence / positive association”. That is support, but short of the chapter’s framing. - Adult ALL, hairy cell leukaemia and myeloproliferative disorders are not established. - The consultants’ narrow position (“only AML”) is also not the mainstream view: IARC keeps several lymphoid outcomes on the positive-association list. So the dispute is unresolved, not won by either side.

Weight. The lesson that the list of recognised outcomes tends to grow over time (digest insight 10) should be graded moderate for this case, as the digest already does. Use the chapter’s disease list as advocacy, not as a statement of consensus.

Claim 3. NCI/CAPM Chinese cohorts show raised risk at average exposures around 1 ppm (pp. 42, 44)#

Original claim. - The Chinese cohort studies show significantly raised relative risks for all lymphatic and blood cancers combined, and for acute non-lymphocytic leukaemia and MDS combined, among workers averaging 1.2 ppm for 5.5 years (6.7 ppm-years). - This result comes from a personal communication (Hayes, 1999, p. 42). - A 1 ppm limit allows 45 ppm-years over a working life (p. 42). - The studies show “high relative risks for leukaemia, myelodysplastic syndrome and non-Hodgkin’s lymphoma” at “very low average benzene exposures, such as around 1 ppm” (p. 44). - Industry consultants’ critiques of the exposure estimates (Wong, 1999; Budinsky et al., 1999) “misrepresent the data” (p. 44).

What happened since.

• Blood toxicity below 1 ppm.
• Lan et al. (2004, Science) reported lower white cell and platelet counts at exposures below 1 ppm, with genetic variants affecting susceptibility.
• It was challenged by Lamm and Grünwald (2006, Science letter). The name S. H. Lamm also appears on the 1977 “random cluster” letter the chapter criticises (p. 43).
• The NCI group’s pooled reanalysis (Vermeulen et al., 2023) found a “supra-linear” exposure–response. The drop in cell counts is proportionally larger at lower exposures, across roughly 0.1–100 ppm.
• Regulators have not fully accepted this. RAC judged the lowest effect level for blood toxicity “in the range of 2 ppm and above”, while listing Lan et al. (2004) among the studies considered. It set its limit on chromosome damage, with a lowest effect level of about 1 ppm, instead.
• An industry consortium’s quality-weighted review put the lowest effect level for blood toxicity near 2 ppm and found no effects at 0.59 ppm (Schnatter et al., 2020).
• Benzene poisoning at low cumulative exposure. In the NCI–China cohort, excess absolute risk of benzene poisoning rose “at low cumulative benzene exposure levels with no threshold”. It was 0.5% in the lowest category (above 0 to 10 ppm-years) (Vermeulen et al., 2022).
• Cancer at low exposure, from independent and industry-linked cohorts.
• Australian petroleum workers: leukaemia risk rose above 2 ppm-years (Glass et al., 2003).
• Pooled nested case-control studies of petroleum workers in Australia, Canada and the UK, industry-affiliated and led by an ExxonMobil epidemiologist:
  • MDS risk rose monotonically, with an odds ratio of 4.33 for more than 2.93 ppm-years versus 0.348 ppm-years or less (Schnatter et al., 2012);
  • the AML risk was “not persuasive” (Rushton et al., 2014).
• Norwegian offshore workers, with maximum average intensity of about 0.04 ppm: dose-related patterns for AML and multiple myeloma (Stenehjem et al., 2015).
• A meta-analysis found raised leukaemia risk even in the lowest cumulative-exposure category, below 40 ppm-years (Khalade et al., 2010).
• Were the exposure critiques upheld?
• IARC continued to treat the NCI–China cohort as a key informative study in 2009 and 2017. In 2009 it excluded a series of consultant-authored meta-analyses of petroleum workers (Wong and Raabe, 1995–2000), citing methodological concerns (IARC Vol. 100F).
• But the NCI group did rebuild the exposure assessment with a more elaborate calibrated model (Portengen et al., 2016). Its partial validation found “a potential downward bias at low (<1 mg/m³) exposure estimates”. That is broadly the direction the critics had argued: underestimating exposure would overstate risk per ppm.
• With the rebuilt estimates, the MDS/AML trend was only suggestive and the NHL trend absent (Linet et al., 2019; Linet et al., 2020).
• So the critiques were not “upheld” as a reason to discard the data. But they were partly borne out as a reason to temper the low-dose claims, especially for lymphoma.
• Low-dose metabolism: a second, unresolved front. The 2013 EEA update (Annex 3, p. 721) added an argument the chapter did not make. It said that below about 1 ppm, benzene is turned into its toxic metabolites more efficiently, so risk assessments based on workers exposed at higher levels “are likely to underestimate risk by 3- to 9-fold”. It cited NCI–Berkeley studies of workers in Tianjin, China (Kim et al., 2006; Rappaport et al., 2009). That dispute is still open:
• Contested by industry-affiliated authors. Consultants writing with scientists from ExxonMobil, Shell, LyondellBasell and the refiners’ association Concawe reanalysed the same Tianjin data.
  • They argued that the apparent low-dose efficiency is a statistical artefact of dividing one noisy measure by another (Cox et al., 2017).
  • They found metabolism “approximately linear” below about 15 ppm, and “no evidence that the dose-response relationship is supra-linear” (Cox et al., 2021).
  • A 2024 analysis co-authored with Shell found that metabolites from sources other than benzene confound biomonitoring studies. It said this “precludes making any assessments of how Bz metabolism differs below approximately 3 ppm” (Hays et al., 2024).
• Maintained by the original group. In 2025 the original group refitted the data. It found evidence for a second, high-affinity pathway (probably in the lung) that dominates at the parts-per-billion levels of everyday air. It concluded that risks predicted from workers exposed above 1 ppm “likely underestimate risks to the general public by many fold” (Thomas et al., 2025). Two of its authors disclose paid consulting and testimony in benzene litigation. One disclosed earlier research funding from the American Petroleum Institute (API) and the American Chemistry Council.
• No regulatory adoption found. I found no regulatory body that has adopted a low-dose metabolic correction. RAC (2018) went the other way, judging a threshold defensible for chromosome damage.

Verdict: partly held up. - Risk of myeloid disease (MDS and AML) at exposures well below a 45 ppm-year working lifetime at 1 ppm is now supported by several independent and industry-affiliated studies. - Blood toxicity below 1 ppm is reported but disputed by regulators. - The specific claims of “high relative risks… for NHL” at around 1 ppm, and the personal-communication figure of 6.7 ppm-years, have not been confirmed by the same group’s later analyses.

Weight. The lesson that uncertainty about past exposures is irreducible (p. 44; digest insight 7) is strongly reinforced. Twenty years later, the investigators rebuilt their own exposure estimates and some results changed. Use the low-dose claims for myeloid disease with confidence and the lymphoid claims with caution. The metabolism dispute shows how one data set can support opposite readings of low-dose risk. The readings split along the lines of who funded or employed the analysts, and both camps declare interests. Treat the 2013 “3- to 9-fold underestimate” as one camp’s hypothesis, not a finding.

Claim 4. The 1977–1987 delay will cause 198 extra leukaemia and 77 extra myeloma deaths (p. 41); Table 4.1 says it “caused more than 200 deaths” (p. 47)#

Original claim. - This is a projection (Infante and DiStasio, 1988, Lancet letter). - It combines OSHA’s final risk assessment with estimated excess exposure among US workers during the delay. It “will eventually develop” 198 leukaemia and 77 multiple myeloma deaths (p. 41). - The EEA-compiled Table 4.1 restates this as a completed fact: “This 10-year delay caused more than 200 deaths” (p. 47).

What happened since. - I found no later study that tried to verify these projected deaths. Given the latency and the lack of individual exposure records, none could. - The multiple myeloma part now rests on weaker ground: - the Pliofilm myeloma excess did not persist in the 2002 update (Rinsky et al., 2002); - IARC rates myeloma as “limited evidence/positive association” (Vols. 100F and 120). - Infante’s 2006 meta-analysis of cohort studies reported a pooled relative risk for myeloma of 2.13 (95% CI 1.31–3.46) (Infante, 2006, Ann N Y Acad Sci). This is his own analysis, and IARC did not upgrade the evaluation. - An independent estimate not cited by the chapter (Nicholson and Landrigan, 1989, Environmental Health Perspectives, pre-2001) put leukaemia deaths from exposures above 1 ppm between 1978 and 1987 at 30 to 490. That range brackets Infante’s 198 but shows the uncertainty is roughly sixteen-fold. - The EEA repeated the past-tense wording in 2013. Annex 2 of Late lessons II (Table A2.4, p. 705) reprints “This 10-year delay caused more than 200 deaths in the United States” unchanged, with no caveat. - A new argument that the underlying risk model understates deaths. The 2013 Annex 3 (p. 721) argued that the Pliofilm cohort, the basis of OSHA’s 1987 risk assessment, misses “approximately 4–9 additional leukemia deaths”. This rested on two papers then “submitted for publication”. - Part I, as published, says a document review shows “between two and five workers diagnosed with [AML] could be added” (Infante, 2013, IJOEH). - Part II, as published, is a short paper on “take-home” leukaemia, not the statistical estimate the annex described. I found no published version of the 4–9 estimate. - I found no agency that has revised a Pliofilm-based risk estimate to include the added cases.

Verdict: unclear. - It is plausible that the delay cost lives, and an independent estimate agrees on the order of magnitude. - The point figures are model outputs that cannot be verified. - The myeloma component has been weakened. - The EEA’s past-tense restatement (“caused”), repeated in 2013, overstates what the evidence supports.

Weight. Use “the delay probably cost tens to hundreds of lives” rather than “caused more than 200 deaths”. The general lesson that delay has a human cost is sound. But this case’s numbers illustrate a second lesson the chapter does not draw: counts of harm from delay depend heavily on the model chosen. They can be stated with more certainty than the model supports, as Table 4.1 does.

Claim 5. The 1980 Benzene Decision made risk assessment a “straightjacket” that adds years of delay (pp. 40–41; echoed by the editors, p. 181)#

Original claim. - After the Supreme Court’s 1980 Benzene Decision, OSHA’s risk analyses became a “straightjacket”. - The analyses are “encumbered” by arguments about mechanism and choice of animal species, so the process “has created additional years of delay in standard-setting” (p. 41). - The Court itself had said the requirement was “not a mathematical straitjacket” (p. 40).

What happened since. - Pace of OSHA standards. The Government Accountability Office (GAO) reviewed OSHA standard-setting from 1981 to 2010 (GAO-12-330, 2 April 2012). - Standards took “from 15 months to 19 years, and averaged more than 7 years” (93 months). - Agency officials said the Benzene Decision “essentially established a standard of medical and scientific certainty”. They said it “resulted in OSHA staff having to spend an inordinate amount of effort gathering data”. - GAO notes that OSHA “generally conducts quantitative risk assessments for each health standard”. - Wholesale updating blocked. A 1992 appeals court decision struck down OSHA’s 1989 attempt to update exposure limits for more than 400 air contaminants at once, for lack of substance-by-substance findings of significant risk (GAO-12-330). OSHA counts only 16 agents with complete new exposure limits set since 1970 (OSHA annotated-PELs page). - Recent health standards. Both took nearly two decades: - crystalline silica: work began in 1997; final rule 81 FR 16285, 25 March 2016; - beryllium: work began in 2000; final rule 82 FR 2470, 9 January 2017.

I found no new OSHA chemical exposure-limit standard since 2017. This is based on OSHA’s own count and the absence of later final rules on its Federal Register pages consulted. It is not an exhaustive docket search. - Evidence that cuts the other way. - GAO found that health standards took less time on average than safety standards: about 6 years 4 months against about 8 years 6 months. - GAO attributes delay to several factors: “increased procedural requirements”, “shifting priorities” and “a rigorous standard of judicial review”. - GAO also lists the Regulatory Flexibility Act, Paperwork Reduction Act, Congressional Review Act, Information Quality Act and OMB review under Executive Order 12866. Most of these requirements date from 1980 or later. - So the Benzene Decision is one strand of a larger procedural burden, not the sole cause. - Mechanism-based analysis is not intrinsically slow or weak. The EU RAC used exactly the kind of mechanism-based reasoning Infante criticised (p. 41), a threshold for chromosome damage. It arrived at a limit 20 times lower than OSHA’s (RAC, 2018). What matters is who carries the burden of proof and how assessment feeds into decisions. - The legal foundations remain contested. In 2024 the Supreme Court declined to hear a challenge to the delegation of power in the OSH Act (the law that created OSHA). Justice Thomas dissented and Justice Gorsuch would have taken the case (Allstates Refractory Contractors v. Su, No. 23-819, cert. denied 2 July 2024).

Verdict: partly held up. - The prediction of continued slow standard-setting was borne out. OSHA officials themselves attribute part of it to the Benzene Decision’s evidentiary demands. - The specific causal claim that risk and mechanism analysis is the source of delay is only partly supported. GAO spreads the blame across many procedural layers, and the EU case shows mechanism-based assessment can support stricter limits.

Weight. Strong support for the general lesson that placing the burden of proof on the regulator, combined with procedural hurdles, produces indefinite inaction (digest insights 6–7; pp. 40–41, 44). Weaker support for blaming quantitative risk assessment in itself.

Claim 6. Industry-consultant critiques of the Chinese studies may repeat the post-1977 delay; governments may use these data for general-population risk (pp. 44–45)#

Original claim. - Consultants’ critiques of the Chinese studies (Wong, 1998, 1999; Budinsky et al., 1999) are “reminiscent of the protracted debate and delay” after 1977 (p. 45). - The studies “may be used in the future by governments in Europe, the United States and other countries for estimating benzene-related diseases from low-level exposure to the general population” (p. 44).

What happened since. - Did governments use the data? - The US EPA’s IRIS database bases benzene’s inhalation reference concentration and oral reference dose (last revised 17 April 2003) on Rothman et al. (1996). That is an NCI/CAPM study of lymphocyte counts in Shanghai workers. This is a general-population value derived from the Chinese research programme. - EPA’s cancer unit risk (2000) still rests on the US Pliofilm cohort. - IARC relied on the NCI–China cohort in both 2009 and 2017. - RAC (2018) cites the Chinese blood-toxicity studies (Lan et al., 2004; Rothman et al., 1996) among those considered, but set its limit on chromosome damage. - California (OEHHA) revised its 8-hour and chronic reference levels to 3 µg/m³ in 2014. I could not retrieve the technical document to confirm which study it relied on. - For the general public, the EU tightened its ambient air limit from 5 µg/m³ to 3.4 µg/m³, to apply by 1 January 2030 (Directive (EU) 2024/2881). - Did contestation continue? Yes, broadly as the chapter anticipated: - The API-funded Chinese programme. Members of the American Petroleum Institute funded the “Shanghai Health Study” (2001–2009) in China (Gross and Paustenbach, 2018). In 2006 Infante alleged that the API “has already announced the intended conclusions” of this research (Infante, 2006, IJOEH). This is an allegation by the chapter’s author, not an independent finding. - The study’s published summary reports raised risks of benzene poisoning and aplastic anaemia “following sufficient exposure to relatively high airborne concentrations”. - It found “NHL was not significantly increased”. - It argued for an inflammation-based mechanism rather than one starting with chromosomal abnormalities. - Consultancy meta-analyses. These found no link with NHL or CML (Alexander et al., 2010, Exponent; Weed, 2010; Lamm et al., 2005, 2009). - Did a comparable delay follow? - In the US, OSHA has not reopened the benzene limit at all. So there was no repeat of the 1977–87 legal battle, only inaction. - In the EU, the limit went from RAC’s opinion (2018) to a binding law (2022) and full application (2026).

Verdict: partly held up. - Both halves of the forecast came true in part: - the data were used by governments, most clearly in EPA’s non-cancer values; - industry-linked contestation continued. - The feared repeat of the post-1977 delay did not happen in the same form. The US never tried again, and the EU acted within about eight years of RAC’s opinion.

Weight. Reinforces the lesson that contestation grows as evidence strengthens and stakes rise (digest insight 5, pp. 43–45). But the post-2001 record adds a qualification: industry-affiliated studies also produced evidence of low-dose risk (MDS), and an industry consortium proposed a limit below 1 ppm. Contestation and convergence can run together.

Claim 7. ACGIH limits were set at “easily achievable” levels under corporate influence; an independent, open international alternative “has not been achieved” (pp. 43, 46); consensus bodies should keep their distance from producers and their consultants (p. 46)#

Original claim. - Consensus organisations “usually base their recommended exposure levels on what is easily achievable” (p. 43). - Castleman and Ziem (1988) found that ACGIH threshold limit values (TLVs) were heavily influenced by corporations. - An international, open effort to replace TLVs “has not been achieved” (p. 43). - Lesson: consensus bodies “should maintain distance from the producers of the chemicals and their ‘consultants’” (p. 46).

What happened since. - ACGIH benzene value. In the 2024 cycle ACGIH lowered its benzene TLV from 0.5 ppm (short-term 2.5 ppm) to 0.02 ppm (ACGIH page; archived captures of 6 December 2023 and 4 April 2024). - This is now 50 times below OSHA’s legal limit, 10 times below the EU binding limit, and below RAC’s recommended 0.05 ppm. - The supporting documentation is behind a paywall, so I could not check its basis. - On benzene, the consensus body is now the most protective standard-setter, not the least. - ACGIH conflict-of-interest policy. - It requires written and oral disclosure of “potential sources of bias and conflict of interest”, including employment by or consultancy for affected organisations. - Members can be asked to resign for “more severe or extensive conflicts”. Failure to disclose leads to dismissal. Imbalance is handled by adding members. - It singles out the threshold-limit committees as “at the core” of risk assessment. - It is a disclosure-and-balance model, not the exclusion of industry-affiliated scientists that the chapter’s lesson implies. - An open, public alternative, in Europe. - The EU’s scientific assessment of occupational limits moved from the Scientific Committee on Occupational Exposure Limits (SCOEL) to ECHA’s RAC. - RAC’s benzene opinion followed a public call for comments (10 October to 7 November 2017) and was published with rapporteurs named. It recommended 0.05 ppm. - Industry interests then entered openly at the tripartite ACSH stage, where the binding value was set at 0.2 ppm, four times RAC’s figure (ACSH opinion, 2019). - In effect, the EU has separated the hazard assessment from the feasibility bargain, the structural fix the chapter’s lesson points toward. That is short of an “international” (global) replacement. - Infante’s later concern moved to IARC. In 2018 he co-authored a commentary defending IARC against campaigns by “those with economic interests” (Infante et al., 2018).

Verdict: partly held up. - The historical diagnosis, that limits were set at achievable levels, is not contradicted. - But the specific target, ACGIH, now sets the strictest benzene value of any major body. It manages conflicts of interest through disclosure and balance rather than distance. - The claim that an open alternative “has not been achieved” is partly overtaken in the EU, though not globally.

Weight. The lesson is better stated in structural, technology-neutral terms. Keep the scientific assessment of hazard separate from, and published before, the negotiation over feasibility. Then make the gap between the two visible: 0.05 ppm versus 0.2 ppm in the EU case (pp. 43, 46). The “consensus bodies are captured” version is dated for this substance.

Claim 8. Benzene in gasoline is a poorly communicated public hazard; pumps should carry warnings (pp. 45–46)#

Original claim. - US gasoline averaged about 1.5% benzene and could reach 5%. European petrol historically contained more, “supposedly reduced more recently” (p. 45). - Pumps and safety data sheets do not inform users. - Warning labels at pumps are recommended. - Failing to inform the public would “repeat our failures of the 20th century in the 21st” (p. 46).

What happened since. - Benzene content fell. - EU: maximum 1.0% by volume from 1 January 2000 (Directive 98/70/EC, Annex I). The chapter’s “supposedly reduced” was in fact a binding cap. - US: EPA’s mobile-source air toxics rule required an annual average of 0.62% by volume from 1 January 2011, and a maximum average of 1.3% from 1 July 2012 (EPA gasoline MSAT page). - The hazard of gasoline itself was upgraded. IARC’s Vol. 138 working group (results announced 21 March 2025) classified automotive gasoline as carcinogenic to humans (Group 1). - It “causes cancer of the urinary bladder and acute myeloid leukaemia in adults”. - Evidence was limited for childhood ALL, NHL (including CLL), multiple myeloma, MDS and stomach and kidney cancers. - In 2026, an API- and fuel-industry-funded consultancy review argued that the genotoxicity seen in fuel station attendants “cannot be directly attributed to automotive gasoline” (Lea et al., 2026). - Warnings. - California’s safe-harbour warning for service stations (27 CCR §25607.26–.27) was filed 30 August 2016 and took effect 30 August 2018. - It requires a sign at each pump, in at least 22-point type, naming benzene. - The sign reads that breathing the air or skin contact “can expose you to chemicals including benzene… known to the State of California to cause cancer and birth defects or other reproductive harm”. - It names cancer in general, not the specific cancers Infante wanted listed. - I found no federal US pump-warning requirement and no EU-level requirement for pump-point cancer warnings. - EU hazard labelling (the CLP Regulation) classifies many gasoline streams as carcinogens (category 1B, H350). But I found no provision addressing warnings at the dispenser. - Workers. OSHA’s benzene standard still excludes the “storage, transportation, distribution, dispensing, sale or use of gasoline… subsequent to its final discharge from bulk wholesale storage facilities”. The exception is indoor dispensing for more than 4 hours a day (29 CFR 1910.1028(a)(2)(i)). - Those workers fall back on the older Table Z-2 limits (10 ppm TWA). - The groups Infante named as at risk (p. 45), including station attendants and mechanics, therefore remain largely outside the 1 ppm rule. - Childhood leukaemia. - In 2009 IARC declined to rely on studies using residential proximity to petrol stations as a stand-in for exposure (Vol. 100F). - In 2017 IARC listed childhood AML as a “positive association” (Vol. 120). - Infante (2017) argued that children may be at higher risk at very low environmental exposure. - The 2013 EEA update went further than the 2001 chapter. Annex 3 (p. 720) argued that gasoline “should be considered a cause of all lymphohematopoietic cancers associated with benzene exposure to both adults and children”. It cited raised childhood leukaemia risk near petrol stations (two French case-control studies) and from parents’ exposure. IARC’s 2025 evaluation matches part of this: Group 1, with sufficient evidence for adult AML. But it found only limited evidence for childhood ALL, NHL, multiple myeloma and MDS. So “all lymphohematopoietic cancers” has not been accepted.

Verdict: strengthened. - The core claim, that gasoline exposure is a real and under-communicated cancer hazard, is strengthened by IARC’s 2025 Group 1 classification. - The communication gap largely persists outside California. - Benzene content has fallen well below the chapter’s figures (about 0.6% in the US, 1% maximum in the EU). This reduces exposure intensity but not the hazard’s existence.

Weight. Strong support for the lesson that diffuse exposure through everyday products outlasts workplace controls (digest insight 11, p. 45), and that regulation of the product can advance while warnings to users lag. The US exemption of retail gasoline workers from the benzene standard is a telling detail.

Claim 9. The 1978 voluntary withdrawal of benzene from US consumer products “has never been adequately validated” (pp. 40, 46)#

Original claim. - Benzene was withdrawn voluntarily in 1978, after a paint stripper was shown to generate up to 200 ppm in a home (p. 40). - The withdrawal was done “by manufacturers on a voluntary basis, and it has never been adequately validated” (p. 46).

What happened since. Some of this is primary material from before 2001 that the chapter did not cite. - The US Consumer Product Safety Commission (CPSC) story. IARC’s 1982 monograph (Vol. 29) records the sequence. - In 1978 the CPSC proposed banning all consumer products (except gasoline and laboratory reagents) containing benzene as an intentional ingredient, or as a contaminant at 0.1% or more. - In 1981 it withdrew the proposal. The basis was that benzene “as currently used in consumer products, did not present a significant risk”. The Commission relied on “data from contacts in industry and information obtained from manufacturers, importers and labellers” (Federal Register, Vol. 46, No. 99, pp. 27910–27911, May 1981). - So the “voluntary” withdrawal took place under a proposed federal ban. It was “validated” mainly by information from industry. That supports Infante’s point. - Current US federal rules. - CPSC treats products with 5% or more benzene by weight as hazardous, requiring special labelling. - Paint solvents with 10% or more require special packaging. - There is no ban (NTP Report on Carcinogens, 15th ed., 2021). - The EU, by contrast, restricts benzene above 0.1% by weight in substances and mixtures placed on the market, with fuel exempted (IARC Vol. 120, section 1.5). - Benzene has kept appearing in consumer products, as a contaminant rather than an ingredient. - Soft drinks, 2005–2007. FDA surveys found benzene formed from benzoate and vitamin C in some drinks. “A small number of products sampled contained more than 5 ppb.” These were reformulated. FDA judged the levels found “do not pose a safety concern”. - Hand sanitisers and sun-care products, 2021. An independent testing laboratory reported benzene of 0.1–16.1 ppm in 17% of hand-sanitiser batches and up to 6.26 ppm in 27% of sun-care batches (as reported by NTP, 2021). - Drug excipients, 2023. In December 2023 FDA issued guidance because certain US Pharmacopeia carbomer monographs “allow for unacceptable levels of benzene”. Carbomers are thickening agents made using benzene. FDA asked the Pharmacopeia to remove the monographs.

• The EEA’s own 2013 summary table leans on the withdrawal. Annex 2 of Late lessons II (Table A2.1, p. 702) gives the benzene case’s “date of effective risk reduction action” as “1978 benzene voluntarily withdrawn from most US consumer products”, counting 81 years of “substantial inaction” from 1897. So the EEA’s headline table uses as its risk-reduction marker the step the chapter says “has never been adequately validated”. It also passes over the 1987 workplace standard, even though the chapter is about workers. This is a small internal inconsistency, but it matters for anyone who quotes the “81 years” figure.

Verdict: held up. - No independent verification of the 1978 withdrawal was found. The regulator’s 1981 decision rested on industry-supplied information. - Post-2001 contamination episodes show that the absence of intentional use did not mean the absence of benzene in consumer products. This is a different route from the one the chapter discussed, and it should be described as such.

Weight. Good support for a technology-neutral lesson: a voluntary withdrawal made under regulatory threat, and verified by the industry’s own reports, is a weak guarantee (pp. 40, 46). Contamination by-products can reintroduce a hazard through routes the original commitment never covered.

Claim 10. In the workplace, “the precautionary principle is not relevant”: the failure was prevention of a known harm, not precaution under uncertainty (p. 46)#

Original claim. - “In the case of benzene exposure in the workplace, the precautionary principle is not relevant” (p. 46). - Recommendations to substitute benzene went unheeded for decades despite high rates of disease.

What happened since. - The framing is sound for the period the chapter documents, 1897 to about 1980. Nothing published since disputes that blood disease was known and exposures stayed far above the levels already linked to harm (pp. 38–39, 42–43). Infante’s later work extends this into concealment: - He reports that in the late 1950s and early 1960s the Pliofilm employer did not tell workers with AML their diagnosis, withheld benzene use from the treating haematologist, and resisted compensation until 1968 (Infante, 2013). - The claims come from Infante’s reading of historical documents and have not been independently checked here. - The post-2001 frontier is a precaution problem. Every live dispute since 2001 is about acting under uncertainty: - whether blood toxicity occurs below 1 ppm (Lan et al., 2004, versus RAC’s lowest effect level of about 2 ppm); - whether low-dose cancer risk is linear, supra-linear or has a threshold (Vermeulen et al., 2023, versus RAC, 2018); - which lymphoid outcomes are causal; - how far gasoline and trace contamination matter.

RAC’s use of assessment factors, including an extra factor for bone-marrow sensitivity and disease severity, to reach 0.05 ppm is a precautionary device in all but name. The chapter itself anticipated this at p. 45: “It will be unfortunate if more precaution is taken with the use of data… than with the protection of populations”. - The author’s own 2013 update is a precaution argument. Infante’s 2013 Annex 3 note (pp. 720–721) is entirely about low-level and environmental exposure: effects below 1 ppm, children exposed in the womb and near petrol stations, disputed low-dose metabolism and possible undercounting in the risk model. These are claims made under uncertainty, several still unresolved (Claims 2, 3 and 8). In practice, the author’s attention moved from prevention to precaution between 2001 and 2013. - How the EEA framed the 2001 cases in 2013. The 2013 summary describes them as cases where policy was made “against a background of scientific uncertainty and ‘surprises’ — and where clear evidence of hazards… was often ignored” (EEA, 2013, Summary, PDF p. 11). The EEA thus combined the prevention framing and the precaution framing, rather than adopting Infante’s distinction.

Verdict: partly held up. - It is correct and important for the historical workplace case, which is primarily a prevention failure. - It no longer describes benzene as a whole. Benzene’s current regulatory questions are about low doses, the shape of the dose–response curve and diffuse exposure. Those are precaution questions.

Weight. Keep the distinction (digest insight 12; p. 46): it is one of the chapter’s most useful analytical contributions. When benzene is used as an example, split it into two lessons: - (a) an example of failure to act on known harm (strong); - (b) a current example of how precaution operates at the low-dose frontier, where expert bodies diverge by a factor of about 50: ACGIH 0.02 ppm, RAC 0.05 ppm, EU 0.2 ppm, industry consortium 0.25 ppm, OSHA 1 ppm.

Implications for the section’s transferable insights#

Stated in technology-neutral terms and keyed to the digest’s list.

Method and access notes#

• Search. For most of this check the web search tool was unavailable because the session’s search budget was spent. It became available in a final pass, which I used to find the 2013 annex and the post-2013 low-dose metabolism papers. Otherwise I worked from primary documents retrieved directly:
• regulator and agency web pages;
• the EU Publications Office repository, for EUR-Lex texts;
• IARC monograph PDFs;
• PubMed records and abstracts through the NCBI E-utilities;
• the Internet Archive, for a blocked ECHA document and to date the ACGIH change.
• Blocked sites. ECHA, CDC/ATSDR and OEHHA’s live pages blocked automated access. The RAC opinion was read from an Internet Archive copy (captured 21 January 2021). OEHHA values come from an archived page (16 November 2019).
• Not verified:
• the evidence basis of ACGIH’s 2024 value (the documentation is behind a paywall);
• the critical study for OEHHA’s 2014 reference levels and ATSDR’s minimal risk levels;
• whether any individual EU member state requires cancer warnings at pumps.
• 2013 EEA annex. The EEA’s current site no longer serves the 2013 annexes. I read Annexes 2 and 3 from an Internet Archive capture (5 October 2013) of the EEA’s own file, “Late lessons II – Annexes (05-2013)”. Page numbers are the 2013 report’s printed numbers. I did not open the project’s local copy of the 2013 report, because this task was limited to the named files and web sources. A reader with the local copy can confirm pp. 702, 705 and 720–721 there.
• Author context. Infante (OSHA at the time of writing) remained an active advocate after 2001, publishing on benzene and myeloma, IARC’s 2009 evaluation, the Pliofilm cohort, gasoline and childhood leukaemia, and industry influence. His later papers are cited here as his views, not as independent confirmation.
• My calculations. Two figures in the Claim 1 risk table are my own conversions, flagged there: the ~1.5–5 per 1,000 derived from EPA’s unit risk, and the ~7 per 1,000 scaled from the German AGS values. They are not agency figures.

Sources#

Regulation, policy and agency documents 1. US OSHA. 29 CFR 1910.1028, Benzene (current text; amendment history to 84 FR 21598, 14 May 2019). https://www.osha.gov/laws-regs/regulations/standardnumber/1910/1910.1028 (accessed 25 Sep 2026) 2. US OSHA. Permissible Exposure Limits: Annotated Tables (introduction; “outdated and inadequate”; 16 agents). https://www.osha.gov/annotated-pels (accessed 25 Sep 2026) 3. US OSHA. Annotated Table Z-1 and Table Z-2 (benzene rows; Z-2 limits for sectors exempt from 1910.1028). https://www.osha.gov/annotated-pels/table-z-1 ; https://www.osha.gov/annotated-pels/table-z-2 (accessed 25 Sep 2026) 4. US GAO. Workplace Safety and Health: Multiple Challenges Lengthen OSHA’s Standard Setting, GAO-12-330, 2 April 2012. https://www.gao.gov/products/gao-12-330 ; PDF https://www.gao.gov/assets/gao-12-330.pdf 5. US OSHA. Occupational Exposure to Respirable Crystalline Silica; Final Rule, 81 FR 16285, 25 March 2016. https://www.osha.gov/laws-regs/federalregister/2016-03-25-1 6. US OSHA. Occupational Exposure to Beryllium; Final Rule, 82 FR 2470, 9 January 2017. https://www.osha.gov/laws-regs/federalregister/2017-01-09 7. Supreme Court of the United States. Allstates Refractory Contractors, LLC v. Su, No. 23-819, cert. denied 2 July 2024 (Thomas, J., dissenting). Docket: https://www.supremecourt.gov/search.aspx?filename=/docket/docketfiles/html/public/23-819.html ; dissent: https://www.supremecourt.gov/opinions/23pdf/23-819_m648.pdf 8. European Parliament and Council. Directive (EU) 2022/431 of 9 March 2022 amending Directive 2004/37/EC (OJ L 88, 16.3.2022, p. 1), recitals 22–23 and 26, Annex. https://eur-lex.europa.eu/eli/dir/2022/431/oj (text retrieved via http://publications.europa.eu/resource/celex/32022L0431) 9. European Commission. Proposal COM(2020) 571 final, 22 September 2020 (impact estimates: 182 leukaemia cases prevented). Retrieved via http://publications.europa.eu/resource/cellar/59619b36-fccf-11ea-b44f-01aa75ed71a1.0007.03/DOC_1 ; procedure file https://oeil.secure.europarl.europa.eu/oeil/popups/ficheprocedure.do?reference=2020/0262(COD)&l=en 10. Advisory Committee on Safety and Health at Work (ACSH). Opinion on an EU binding occupational exposure limit value for benzene, Doc. 1056/19, adopted 4 June 2019. https://circabc.europa.eu/sd/a/b28832c6-8cc6-4a6c-b966-986211b180fc/Doc.1056-19-EN-ACSH%20CMD_Opinion_benzene%20Adoped%2004062019.pdf 11. ECHA Committee for Risk Assessment (RAC). Opinion on scientific evaluation of occupational exposure limits for benzene, ECHA/RAC/O-000000-1412-86-187/F, adopted 9 March 2018. https://echa.europa.eu/documents/10162/13641/benzene_opinion_en.pdf/4fec9aac-9ed5-2aae-7b70-5226705358c7 (read via Internet Archive capture of 21 Jan 2021) 12. EU-OSHA. Directive 2004/37/EC, carcinogens, mutagens or reprotoxic substances at work (amendment history). https://osha.europa.eu/en/legislation/directives/directive-2004-37-ec-carcinogens-or-mutagens-at-work (accessed 25 Sep 2026) 13. European Parliament and Council. Directive 98/70/EC relating to the quality of petrol and diesel fuels (OJ L 350, 28.12.1998, p. 58), Annex I (benzene max 1.0% v/v). https://eur-lex.europa.eu/eli/dir/1998/70/oj 14. European Parliament and Council. Directive (EU) 2024/2881 on ambient air quality (recast), Annex I (benzene 3.4 µg/m³ by 1 January 2030). https://eur-lex.europa.eu/eli/dir/2024/2881/oj 15. ACGIH. Benzene substance page (TLV-TWA 0.02 ppm; Skin; A1). https://www.acgih.org/benzene-2/ (accessed 25 Sep 2026). Archived earlier values: https://web.archive.org/web/20231206151302/https://www.acgih.org/benzene-2/ (0.5/2.5 ppm) and https://web.archive.org/web/20240404214025/https://www.acgih.org/benzene-2/ (0.02 ppm) 16. ACGIH. Conflict of Interest Policy. https://www.acgih.org/science/tlv-bei-guidelines/policies-procedures-presentations/conflict-of-interest-policy/ (accessed 25 Sep 2026) 17. US EPA. IRIS Chemical Assessment Summary: Benzene (RfC/RfD last revised 17 April 2003; carcinogenicity 19 January 2000). https://iris.epa.gov/static/pdfs/0276_summary.pdf 18. US EPA. Gasoline Mobile Source Air Toxics (0.62 vol% annual average from 1 January 2011; 1.3 vol% maximum average from 1 July 2012). https://www.epa.gov/gasoline-standards/gasoline-mobile-source-air-toxics (accessed 25 Sep 2026) 19. California Code of Regulations, Title 27, §25607.26 and §25607.27, service station warnings (filed 30 August 2016; operative 30 August 2018). https://www.law.cornell.edu/regulations/california/27-CCR-25607.26 ; https://www.law.cornell.edu/regulations/california/27-CCR-25607.27 20. California OEHHA. Benzene chemical page (8-hour and chronic RELs 3 µg/m³, revised 2014; Prop 65 listings 1987 and 1997). Archived 16 Nov 2019: https://web.archive.org/web/20191116213727/https://oehha.ca.gov/chemicals/benzene 21. US National Toxicology Program. Report on Carcinogens, 15th ed. (2021), Benzene profile. https://ntp.niehs.nih.gov/sites/default/files/ntp/roc/content/profiles/benzene.pdf 22. US FDA. Reformulating Drug Products That Contain Carbomers Manufactured With Benzene, guidance, December 2023 (docket FDA-2023-D-5408). https://www.fda.gov/regulatory-information/search-fda-guidance-documents/reformulating-drug-products-contain-carbomers-manufactured-benzene 23. US FDA. Questions and Answers on the Occurrence of Benzene in Soft Drinks and Other Beverages (2005–2007 survey). https://www.fda.gov/food/environmental-contaminants-food/questions-and-answers-occurrence-benzene-soft-drinks-and-other-beverages (accessed 25 Sep 2026) 24. European Environment Agency. Late lessons from early warnings: science, precaution, innovation — Summary, EEA Report No 1/2013. https://www.eea.europa.eu/en/analysis/publications/late-lessons-2 24a. European Environment Agency. Late lessons from early warnings: science, precaution, innovation, EEA Report No 1/2013, Annexes (05-2013): Annex 2, Tables A2.1 (p. 702) and A2.4 (p. 705); Annex 3, Infante PF, “Benzene and gasoline: an update on risk of adult and childhood cancers and issues with quantitative risk assessment” (pp. 720–721). Archived EEA file (captured 5 Oct 2013): https://web.archive.org/web/20131005100848/http://www.eea.europa.eu/publications/late-lessons-2/late-lessons-2-full-report/late-lessons-2-annexes (original URL http://www.eea.europa.eu/publications/late-lessons-2/late-lessons-2-full-report/late-lessons-2-annexes)

IARC evaluations

25. IARC. Monographs Vol. 29, Some Industrial Chemicals and Dyestuffs (1982), benzene (CPSC 1978 proposed ban; 1981 withdrawal, 46 FR 27910–27911). https://publications.iarc.who.int/47 ; PDF https://publications.iarc.who.int/download/mono29.pdf
26. IARC. Monographs Vol. 100F (meeting October 2009; published 2012), Benzene. https://publications.iarc.who.int/123 ; chapter PDF https://publications.iarc.who.int/download/mono100F-24_new.pdf
27. IARC. Monographs Vol. 120, Benzene (meeting 10–17 October 2017; published 2018), Sections 1, 5 and 6. https://publications.iarc.who.int/576 ; https://publications.iarc.who.int/download/120-12-Section6.pdf ; https://publications.iarc.who.int/download/120-11-Section5.pdf ; https://publications.iarc.who.int/download/120-07-Section1.pdf
28. IARC. News, “Volume 138: Automotive gasoline and some oxygenated gasoline additives”, 21 March 2025. https://monographs.iarc.who.int/news-events/volume-138-automotive-gasoline-and-some-oxygenated-gasoline-additives/ ; Turner MC et al., Lancet Oncology, published online 21 March 2025. https://doi.org/10.1016/S1470-2045(25)00165-2

Peer-reviewed studies and commentaries

29. Lan Q et al. Hematotoxicity in workers exposed to low levels of benzene. Science 2004;306:1774–6. https://doi.org/10.1126/science.1102443
30. Lamm SH, Grünwald HW. Benzene exposure and hematotoxicity (letter). Science 2006;312:998–9. https://doi.org/10.1126/science.312.5776.998b
31. Vermeulen R et al. Nonlinear low dose hematotoxicity of benzene; a pooled analyses of two studies among Chinese exposed workers. Environ Int 2023;177:108007. https://doi.org/10.1016/j.envint.2023.108007
32. Vermeulen R et al. Benzene exposure and risk of benzene poisoning in Chinese workers. Occup Environ Med 2022;79:610–17. https://doi.org/10.1136/oemed-2021-108155
33. Linet MS et al. A retrospective cohort study of cause-specific mortality and incidence of hematopoietic malignancies in Chinese benzene-exposed workers. Int J Cancer 2015;137:2184–97. https://doi.org/10.1002/ijc.29591
34. Portengen L et al. Retrospective benzene exposure assessment for a multi-center case-cohort study of benzene-exposed workers in China. J Expo Sci Environ Epidemiol 2016;26:334–40. https://doi.org/10.1038/jes.2015.44
35. Linet MS et al. Benzene exposure response and risk of myeloid neoplasms in Chinese workers: a multicenter case-cohort study. J Natl Cancer Inst 2019;111:465–74. https://doi.org/10.1093/jnci/djy143
36. Linet MS et al. Benzene exposure-response and risk of lymphoid neoplasms in Chinese workers: a multicenter case-cohort study. Am J Ind Med 2020;63:741–54. https://doi.org/10.1002/ajim.23142
37. Bassig BA et al. Occupational exposure to benzene and non-Hodgkin lymphoma in a population-based cohort: the Shanghai Women’s Health Study. Environ Health Perspect 2015;123:971–7. https://doi.org/10.1289/ehp.1408307
38. DeMoulin D et al. Occupational benzene exposure and cancer risk among Chinese men: the Shanghai Men’s Health Study. Cancer Epidemiol Biomarkers Prev 2024;33:1465–74. https://doi.org/10.1158/1055-9965.EPI-24-0325
39. Glass DC et al. Leukemia risk associated with low-level benzene exposure. Epidemiology 2003;14:569–77. https://doi.org/10.1097/01.ede.0000082001.05563.e0
40. Schnatter AR et al. Myelodysplastic syndrome and benzene exposure among petroleum workers: an international pooled analysis. J Natl Cancer Inst 2012;104:1724–37. https://doi.org/10.1093/jnci/djs411
41. Rushton L et al. Acute myeloid and chronic lymphoid leukaemias and exposure to low-level benzene among petroleum workers. Br J Cancer 2014;110:783–7. https://doi.org/10.1038/bjc.2013.780
42. Glass DC et al. Risk of myeloproliferative disease and chronic myeloid leukaemia following exposure to low-level benzene. Occup Environ Med 2014;71:266–74. https://doi.org/10.1136/oemed-2013-101664
43. Stenehjem JS et al. Benzene exposure and risk of lymphohaematopoietic cancers in 25 000 offshore oil industry workers. Br J Cancer 2015;112:1603–12. https://doi.org/10.1038/bjc.2015.108
44. Khalade A et al. Exposure to benzene at work and the risk of leukemia: a systematic review and meta-analysis. Environ Health 2010;9:31. https://doi.org/10.1186/1476-069X-9-31
45. Vlaanderen J et al. Occupational benzene exposure and the risk of lymphoma subtypes: a meta-analysis of cohort studies. Environ Health Perspect 2011;119:159–67. https://doi.org/10.1289/ehp.1002318
46. Rana I et al. Benzene exposure and non-Hodgkin lymphoma: a systematic review and meta-analysis of human studies. Lancet Planet Health 2021;5:e633–43. https://doi.org/10.1016/S2542-5196(21)00149-2
47. Alexander DD et al. Benzene exposure and non-Hodgkin lymphoma: a meta-analysis of epidemiologic studies. J Occup Environ Med 2010;52:169–89. https://doi.org/10.1097/JOM.0b013e3181cc9cf0
48. Weed DL. Meta-analysis and causal inference: a case study of benzene and non-Hodgkin lymphoma. Ann Epidemiol 2010;20:347–55. https://doi.org/10.1016/j.annepidem.2010.02.001
49. Lamm SH et al. Non-Hodgkin lymphoma and benzene exposure: a systematic literature review. Chem Biol Interact 2005;153–154:231–7. https://doi.org/10.1016/j.cbi.2005.03.027 ; Lamm SH et al. Chronic myelogenous leukemia and benzene exposure: a systematic review and meta-analysis of the case-control literature. Chem Biol Interact 2009;182:93–7. https://doi.org/10.1016/j.cbi.2009.08.010
50. Gross SA, Paustenbach DJ. Shanghai Health Study (2001–2009): what was learned about benzene health effects? Crit Rev Toxicol 2018;48:217–51. https://doi.org/10.1080/10408444.2017.1401581
51. Schnatter AR et al. Derivation of an occupational exposure limit for benzene using epidemiological study quality assessment tools. Toxicol Lett 2020;334:117–44. https://doi.org/10.1016/j.toxlet.2020.05.036
52. Rinsky RA et al. Benzene exposure and hematopoietic mortality: a long-term epidemiologic risk assessment. Am J Ind Med 2002;42:474–80. https://doi.org/10.1002/ajim.10138
53. Nicholson WJ, Landrigan PJ. Quantitative assessment of lives lost due to delay in the regulation of occupational exposure to benzene. Environ Health Perspect 1989;82:185–8 (pre-2001; not cited by the chapter). https://doi.org/10.1289/ehp.8982185
54. Infante PF. The past suppression of industry knowledge of the toxicity of benzene to humans and potential bias in future benzene research. Int J Occup Environ Health 2006;12:268–72. https://doi.org/10.1179/oeh.2006.12.3.268
55. Infante PF. Benzene exposure and multiple myeloma: a detailed meta-analysis of benzene cohort studies. Ann N Y Acad Sci 2006;1076:90–109. https://doi.org/10.1196/annals.1371.081
56. Infante PF. The IARC October 2009 evaluation of benzene carcinogenicity was incomplete and needs to be reconsidered. Am J Ind Med 2011;54:157–64. https://doi.org/10.1002/ajim.20887
57. Infante PF. Benzene and leukemia, Pliofilm revisited: I. Int J Occup Environ Health 2013;19:215–22. https://doi.org/10.1179/2049396713Y.0000000029
58. Infante PF. Residential proximity to gasoline stations and risk of childhood leukemia. Am J Epidemiol 2017;185:1–4. https://doi.org/10.1093/aje/kww130
59. Infante PF, Melnick R, Vainio H, Huff J. Commentary: IARC Monographs Program and public health under siege by corporate interests. Am J Ind Med 2018;61:277–81. https://doi.org/10.1002/ajim.22811
60. Lea IA et al. Assessment of the carcinogenic potential of automotive gasoline in humans based on mechanistic evidence. Curr Res Toxicol 2026;10:100284 (funding: American Petroleum Institute; American Fuel & Petrochemical Manufacturers). https://doi.org/10.1016/j.crtox.2026.100284
61. Infante PF. Benzene and leukemia, Pliofilm revisited: II. Take-home leukemia. Int J Occup Environ Health 2013;19:245–7. https://doi.org/10.1179/2049396713Y.0000000032
62. Kim S et al. Using urinary biomarkers to elucidate dose-related patterns of human benzene metabolism. Carcinogenesis 2006;27:772–81. https://doi.org/10.1093/carcin/bgi297
63. Rappaport SM et al. Evidence that humans metabolize benzene via two pathways. Environ Health Perspect 2009;117:946–52. https://doi.org/10.1289/ehp.0800510
64. Cox LA, Schnatter AR, Boogaard PJ, Banton M, Ketelslegers HB. Non-parametric estimation of low-concentration benzene metabolism. Chem Biol Interact 2017;278:242–55 (authors affiliated with ExxonMobil (retired), Shell, LyondellBasell, Concawe). https://doi.org/10.1016/j.cbi.2017.08.007
65. Cox LA, Ketelslegers HB, Lewis RJ. The shape of low-concentration dose–response functions for benzene: implications for human health risk assessment. Crit Rev Toxicol 2021;51:95–116 (Concawe, ExxonMobil affiliations). https://doi.org/10.1080/10408444.2020.1860903
66. Hays SM, Kirman CR, Cox LA, Sarang SS. Benzene metabolism and health risk evaluation: insights gained from biomonitoring. Crit Rev Toxicol 2024;54:685–93 (Shell co-author). https://doi.org/10.1080/10408444.2024.2379896
67. Thomas R, Kim S, Lan Q, Vermeulen R, Zhang L, Rothman N, Smith MT, Rappaport SM. Benzene metabolism is dominated by a high-affinity pathway at ambient exposures with implications for cancer risks. Int J Mol Sci 2025;26:8550 (published 3 September 2025; authors disclose litigation consulting and prior API/ACC research support). https://doi.org/10.3390/ijms26178550