Atrazine: The Herbicide in Your Water Supply

Atrazine is the most commonly detected pesticide contaminant of ground, surface, and drinking water in the United States Hayes et al. 2010. Concentrations in mid-western tap supplies peak in exactly the weeks when ice trays get refilled most. An EWGEnvironmental Working Group, a US non-profit that publishes drinking-water analyses based on utility-reported data analysis of EPA monitoring records found seasonal spikes running 3 to 7× EWG analysis of EPA 2017 atrazine monitoring data — peak concentrations during late-spring and early-summer runoff in some midwestern systems, calculated against the EPA 3 µg/L MCL the federal legal limit in some systems during late-spring and early-summer runoff. None of those spikes triggered a violation. The legal limit is a running annual averageEPA Safe Drinking Water Act compliance for atrazine is calculated under 40 CFR 141.24 as a running average of quarterly samples — a 22 µg/L peak in June can be diluted across three quiet quarters and the system still reports compliance. (40 CFR 141.24), which means a utility can dilute a 22 µg/L peak in June across three quiet quarters and report compliance for the year.

The ice in the freezer was made out of the spike. This article walks through what atrazine is, why it sits in your water, what the evidence — animal, mechanistic, human — actually shows it does, why the EU banned it twenty-two years ago and the US still permits it, and what an under-£40 fix at the kitchen sink looks like. Part of our evidence-based guide to chemical exposure through tap water, with the broader endocrine-disruptor context here.

Atrazine is an s-triazineA class of nitrogen-containing heterocyclic herbicides — three nitrogen and three carbon atoms in a six-membered ring — that work by inhibiting plant photosynthesis at photosystem II. The triazine ring makes them stable in soil and persistent in water. herbicide that controls broadleaf and grassy weeds in corn, sorghum, sugarcane, and a handful of other crops by blocking photosynthesis at photosystem II — the leaves yellow, the plant starves, the corn underneath grows uncontested. C₈H₁₄ClN₅, CASChemical Abstracts Service 1912-24-9. It was patented by the Swiss firm Geigy in 1958 and is now manufactured globally; Syngenta, headquartered in Basel, is the largest producer.

In the United States, atrazine was applied to 65% USDA NASS 2021 Agricultural Chemical Use Highlights — atrazine applied to 65% of US planted corn acres, the most widely used corn herbicide that year; mesotrione second at 47%, glyphosate isopropylamine salt third at 41% of planted corn acres in 2021 — the most widely used corn herbicide, ahead of mesotrione (47%) and glyphosate isopropylamine salt (41%) USDA NASS 2021. Roughly 30 million pounds of active ingredient leave farm sprayers in a typical year. Most of it goes onto soil between April and June; rain washes a fraction into surface water and groundwater between May and July. The molecule's water solubility (33 mg/L at 22°C) and soil half-life (60-100 days) are the engineering reasons it ends up in the rivers utilities draw from.

Two thirds of US public drinking water comes from surface water — the rivers, lakes, and reservoirs that drain agricultural watersheds. The USGSUnited States Geological Survey National Water-Quality Assessment programme has tracked atrazine in those watersheds since the 1990s. Across the most recent groundwater cycle (2013–2018), atrazine and four of its degradation products were each detected in more than 5% of the public-supply wells sampled — the highest detection profile of any pesticide group surveyed.

EWG's analysis of utility-reported testing data, published in 2018, identified atrazine in the tap water of nearly 30 million Americans EWG 2018 analysis of EPA Consumer Confidence Reports and SDWIS data — atrazine detected in supplies serving roughly 30 million people across 28 states, concentrated in corn-producing watersheds and downstream water systems across 28 states EWG 2018 — number of US states with at least one public water system reporting atrazine detection in the 2015 reporting year, concentrated in the corn belt — Iowa, Illinois, Ohio, Indiana, Missouri, Nebraska, Kansas — and downstream of it. The figure is a synthesis of EPA monitoring data, not an EPA report itself; the underlying utility records are the primary documentation.

Compliance is calculated against a Maximum Contaminant LevelEPA's enforceable federal standard for atrazine is 0.003 mg/L (3 µg/L, often written as 3 ppb), set under 40 CFR 141.61(c) as part of the Phase II rule in 1991 and unchanged since. The MCL is the annual-average ceiling at any given monitoring location; short-term peaks during runoff season can exceed it without triggering a violation if the running annual average stays below. of US 3 µg/L (annual average) 1991 — the EPA's 1991 Phase II standard, set at 0.003 mg/L and not revisited since 40 CFR 141.61. Three parts per billion is roughly a teaspoon and a half of atrazine dispersed across an Olympic-sized swimming pool — small in absolute terms but, on the evidence further down, not biologically inert. The detail that matters is what the standard does and doesn't catch. The MCL is enforced as an annual running average. A utility whose June sample comes back at 22 µg/L can drop to 1 µg/L through the autumn and winter months and report annual compliance — its customers drink the spike whether or not the average crosses the line. The number on the regulator's chart is the year. The number in the glass is the day.

Drinking is the obvious route. It is not the only one. Atrazine in tap water also reaches the body through every other use of household water — and three of atrazine's four common breakdown products are not regulated under the Safe Drinking Water Act at all.

The degradate footnote earns its row. DEADeethylatrazine — atrazine with one ethyl group removed, the most abundant atrazine breakdown product in groundwater, DIADeisopropylatrazine — atrazine with the isopropyl group removed; also commonly detected, and DDADidealkylatrazine — atrazine with both alkyl groups removed; the most stable breakdown product in groundwater share the parent compound's triazine ring and most of its biological activity at the receptor level. None has a federal MCL. Annual utility reports name atrazine but rarely name what atrazine has become by the time it reaches the kitchen tap.

The mechanism that links a corn herbicide to a sex hormone runs through a single enzyme. AromataseCytochrome P450 19A1 — the enzyme that catalyses the final step of estrogen synthesis by converting testosterone to estradiol. Expressed in ovary, testis, brain, fat, and the placenta. Activity tightly regulated; chronic upregulation distorts the testosterone-to-estrogen ratio across tissues. (CYP19A1) is the enzyme that converts testosterone into estrogen — every drop of estradiol your body makes goes through it. Atrazine doesn't bind estrogen receptors directly. It increases the amount of aromatase the cell produces. The mechanism was worked out by Fan and colleagues at Kyushu University in 2007 in human adrenocortical (H295R) cells: atrazine activates the SF-1Steroidogenic Factor 1 — a transcription factor that drives expression of steroidogenic enzymes including aromatase-dependent promoter II that controls aromatase transcription, with knockdown of SF-1 abolishing the effect Fan et al. 2007. The same machinery your endocrine system uses to make sex hormones, switched on harder.

The most-cited animal evidence comes from a Berkeley lab. In 2002, Tyrone Hayes published in PNAS a study exposing African clawed frogs (Xenopus laevisA South African aquatic frog used in developmental biology since the 1930s — well-characterised genetics, transparent embryos, robust in laboratory culture. The standard amphibian model for endocrine and developmental research.) to atrazine concentrations spanning 0.01 to 200 ppb across their entire larval development. At 0.1 ppb Hayes et al. 2002 PNAS — lowest atrazine concentration that produced gonadal abnormality (hermaphroditism) in Xenopus laevis tadpoles, 30× below the EPA drinking water MCL of 3 ppb, thirty times below the EPA's drinking-water limit, exposed males developed testicular oogenesis — eggs growing inside the testes — and demasculinised laryngeal anatomy. At 25 ppb, sexually mature males showed plasma testosterone reduced roughly tenfold relative to controls Hayes et al. 2002. Hayes followed up in 2003 with a leopard frog study that combined the laboratory finding with field-collected animals from atrazine-contaminated US sites — same hermaphroditism, in the wild Hayes et al. 2003.

The 2010 paper is the one that made it into popular culture. Hayes' group exposed male Xenopus to 2.5 ppb Hayes et al. 2010 PNAS — chronic atrazine exposure concentration throughout larval development and up to 3 years post-metamorphosis; below the EPA drinking water MCL of 3 ppb atrazine throughout larval development and for up to three years after metamorphosis — a low, chronic exposure designed to mimic what frogs in atrazine-contaminated ponds actually receive. Among the genetically male animals exposed: 10% Hayes et al. 2010 PNAS — percent of genetic male Xenopus laevis that became functional females after chronic 2.5 ppb atrazine exposure, copulating with control males and producing viable eggs developed into functional females that mated with control males and produced viable eggs Hayes et al. 2010. The remaining exposed males showed reduced breeding gland size, demasculinised larynx, suppressed mating behaviour, decreased spermatogenesis, and significantly reduced plasma testosterone (analysis of variance P<0.025). The frog meme has the science backwards. Atrazine doesn't change orientation. It changes sex.

Industry-affiliated replications report nothing of the sort. Kloas and colleagues in 2009, working under contract through Wildlife International, exposed Xenopus to 0.01–100 µg/L atrazine and reported no effect on growth, larval development, or sexual differentiation Kloas et al. 2009. A parallel Coady study in 2005, with author affiliations spanning the Solomon-Giesy-Kendall industry-aligned cluster, similarly found no effect on metamorphosis, gonadal development, laryngeal muscle, or aromatase activity in Xenopus across 0.1–25 µg/L Coady et al. 2005. The discordance between Hayes' work and the industry-funded replications is the central methodological dispute in atrazine toxicology — and the reason this article has its own H2 below.

The mammalian story runs parallel. Cooper and colleagues at the EPA reported in 2000 that atrazine disrupts the hypothalamic control of the female rat reproductive axis: estrogen-induced luteinizing hormone and prolactin surges were suppressed, and seven of nine proestrus rats developed pseudopregnancy under atrazine exposure Cooper et al. 2000. A 2025 systematic review by Guimarães-Ervilha and colleagues pulled twenty-five rodent studies and ran a meta-analysis on twenty: atrazine reduced FSH, LH, and testosterone (intratesticular and serum) and raised estradiol and progesterone, with effects driven by doses at or above 100 mg/kg Guimarães-Ervilha et al. 2025. The authors' own caveat lands the honest note: "none of the studies have tested doses relevant to human health risk." Murine doses of 100 mg/kg are not what reaches a kitchen tap. The mechanistic picture they paint is.

Human atrazine epidemiology runs through two channels — populations exposed through drinking water (women in pregnancy, with birth-outcome endpoints) and populations exposed occupationally (pesticide applicators in the Agricultural Health Study, with cancer endpoints). Each tells a partial story. Neither is conclusive on its own. The picture they make together is suggestive.

The drinking-water signal began with Iowa. Munger and colleagues at the University of Iowa in 1997 examined intrauterine growth retardation across 856 Iowa municipal water supplies, comparing women in the Rathbun Reservoir communities (high atrazine, multiple pesticide co-exposure) against women on other surface-water supplies. The Rathbun-versus-other relative risk for IUGR came in at 1.8 Munger et al. 1997 EHP — relative risk for intrauterine growth retardation in Iowa Rathbun communities (high atrazine) vs other surface-water communities, 95% CI 1.3–2.7; ecological design with metolachlor and cyanazine co-exposure (95% CI 1.3–2.7) Munger et al. 1997. The authors flagged the ecological design honestly: "a strong causal relationship between any specific water contaminant and risk of IUGR cannot yet be inferred." Atrazine doesn't travel alone in corn-belt water; metolachlor and cyanazine come with it.

The Indiana follow-up tightened the design. Ochoa-Acuña and colleagues at Purdue analysed 24,154 Indiana births against utility-level atrazine measurements and reported a 14% increase Ochoa-Acuña et al. 2009 EHP — adjusted prevalence ratio for small-for-gestational-age birth at gestational atrazine concentrations above 0.644 µg/L mean; 95% CI 1.03–1.24, sample 24,154 Indiana births in small-for-gestational-age prevalence at atrazine concentrations above 0.644 µg/L mean exposure across pregnancy (adjusted prevalence ratio 1.14, 95% CI 1.03–1.24) Ochoa-Acuña et al. 2009. France's PELAGIE birth cohort followed in 2011: Chevrier and colleagues measured urinary atrazine metabolites in 579 pregnant women and reported a 50% increase Chevrier et al. 2011 EHP — odds ratio 1.5 for fetal growth restriction in women with quantifiable urinary atrazine or metabolites; 95% CI 1.0–2.2, PELAGIE birth cohort, n=579 in fetal growth restriction (OR 1.5, 95% CI 1.0–2.2) and a 70% increase in small head circumference for sex and gestational age (OR 1.7, 95% CI 1.0–2.7) among women with detectable atrazine or metabolites Chevrier et al. 2011.

The dose-response curve was pinned by Ohio. A University of Illinois Chicago team linked 14,445 singleton live births to twenty-two community water systems with elevated atrazine, calculating each woman's gestational exposure from her residence's monthly utility records. Term low birth weight rose by 27% Almberg et al. 2018 IJERPH — adjusted odds ratio for term low birth weight per 1 µg/L increase in mean gestational atrazine exposure, 95% CI 1.10–1.45, sample 14,445 Ohio singleton live births per 1 µg/L increase in mean gestational atrazine — adjusted odds ratio 1.27 (95% CI 1.10–1.45). Only 4% Almberg et al. 2018 IJERPH — fraction of monthly utility samples in study water systems that exceeded the EPA 3 µg/L atrazine MCL; the study population was therefore mostly exposed at levels the regulator considered compliant of monthly samples in those Ohio systems exceeded the 3 µg/L MCL Almberg et al. 2018. The exposure was, by regulatory definition, compliant water. The signal was in birth weights anyway — the dose-response gradient running well below the limit.

The occupational channel is the AHSAgricultural Health Study — a long-running prospective cohort of US pesticide applicators and their spouses, primarily in Iowa and North Carolina, established in the early 1990s and a primary source for individual-pesticide cancer epidemiology. Beane Freeman and colleagues' 2011 analysis of 53,662 applicators (36,357 of whom used atrazine) flagged a thyroid cancer signal in the highest exposure quartile Beane Freeman et al. 2011 — though the authors' own caveat is the part to read carefully: "there was no consistent evidence of an association between atrazine use and any cancer site. There was a suggestion of increased risk of thyroid cancer, but these results are based on relatively small numbers and minimal supporting evidence."

The 2024 update — extended follow-up, 8,915 incident cancers in 53,562 applicators — is the honesty hinge. The thyroid signal did not replicate. What replaced it was a different set of suggestive associations: lung cancer relative risk 1.24 (95% CI 1.04–1.46); kidney cancer with a 25-year exposure lag, RR 1.62 (1.15–2.29); aggressive prostate cancer in men under 60, RR 3.04 (1.61–5.75); non-Hodgkin lymphoma in those under 50, RR 2.43 (1.10–5.38) Remigio et al. 2024. None of these are definitive on their own. The pattern across endpoints — birth outcomes in mothers, multiple cancer sites in applicators, consistent reproductive disruption in animals — is what the regulator weighs.

From 1997 until around 2000, Hayes was funded by Syngenta through an external review panel called Ecorisk, set up to study atrazine's amphibian effects. He left the panel after disagreements about publishing his findings and went public with the 2002 PNAS paper. What happened next is documented primarily through Rachel Aviv's reporting in the New Yorker in 2014 and through internal Syngenta materials released via discovery in the Holiday Shores Sanitary District class-action litigation.

The notebook entries of Sherry Ford, a Syngenta communications manager, became part of the public record through the litigation. They include strategy notes such as "discredit Hayes," "prevent citing of TH data by revealing him as noncredible," and "exploit Hayes' faults/problems" Aviv 2014 New Yorker. Aviv's reporting also documents the channels through which the strategy was pursued: ethics complaints to UC Berkeley, requests for Hayes' research data, communications with journals, and surveillance of his lectures. Industry-affiliated replications — Kloas 2009, Coady 2005 — published consistently null findings, and have been cited in regulatory submissions arguing against the relevance of Hayes' work.

The article position is straightforward. Hayes' findings have been replicated in independent labs and are consistent with the in-vitro mechanistic work on aromatase induction. The industry-affiliated null findings exist and should be acknowledged. The reader can hold both at once: an experimental signal across multiple amphibian models, supported by mechanism, contested by manufacturer-sponsored replications run by contract laboratories. The reasonable response when independent academic findings disagree with industry-funded replications is to lean toward the independent academic findings, not to call the question open.

Five jurisdictions, two opposite conclusions on the same evidence base. The EU declined to renew atrazine's approval more than two decades ago. The US is mid-way through a re-evaluation that has been running since 2013.

Two of those rows do most of the work. The EU's 2004 non-inclusion was driven by atrazine contamination of European groundwater at concentrations the regulator considered unacceptable for a precautionary withdrawal. The US 1991 MCL has not moved in thirty-five years. The 2024 EPA mitigation proposal addresses ecological risk to aquatic plants, not the human drinking-water standard — that 3 µg/L number is the same number the agency wrote down before Hayes ever ran his first frog experiment. Twenty-two years of evidence accumulation. The regulatory ceiling is identical to what it was when the first hint arrived.

The cheapest mitigations work, and they sit between £30 and £400 depending on how much of the household water you want to address. The kitchen tap is the right starting point — that's where the dose lands. This is the eso-friendly response when the evidence is strong but the regulatory ceiling is high: switch where switching is cheap, regardless of where the limit currently sits.

What none of these do is address atrazine in food or atrazine in the broader environment. The kitchen-tap fix is a kitchen-tap fix. For most people in atrazine-impacted water systems, that's where the dose is — drinking water exposure dominates dietary exposure for atrazine in a way that's not true for, say, organochlorine pesticides. A KDF-55A high-purity copper-zinc alloy filter medium that addresses chlorine and some heavy metals, used as a pre-filter stage. Pairs well with downstream activated carbon for atrazine and other organic contaminants. pre-filter plus solid-block carbon at point-of-use covers chlorine, atrazine, and most other organic contaminants in one cartridge. A reverse osmosis system covers everything else. Pick the level of protection the water in your zone actually warrants — and let the report tell you which it is.

Atrazine is the most commonly detected pesticide in US drinking water and the most widely used corn herbicide in the country. The animal evidence is consistent across independent labs and supported by an in-vitro mechanism — aromatase induction, the same enzyme your body uses to make estrogen. The human evidence runs thinner but covers reproductive endpoints in pregnant women and multiple cancer signals in applicators. The EU banned it twenty-two years ago. The US permits it with a regulatory ceiling that has not moved since 1991 and an enforcement framework that hides summer peaks inside annual averages.

The next step is a £30 carbon filter and a five-minute look at your utility's water quality report. The ice in the freezer was made out of last summer's runoff. Most of the chemistry that gets into water gets there because nobody priced filtering it out. The household can — at the kitchen tap, for less than the price of a takeaway. The regulator's chart shows a year. The glass on the counter shows a day.