Ask whether Roundup causes cancer and you will hear two confident answers. Bayer says regulators worldwide have found glyphosate safe. Plaintiffs say the World Health Organization calls it a probable carcinogen. Both statements are technically true, which is exactly why the question confuses people. The real answer lives in the epidemiology — the studies of actual humans exposed to glyphosate — and that body of evidence is neither a clean exoneration nor a slam dunk. It is a genuine scientific disagreement with a discernible weight of evidence. This piece walks through the main studies one at a time, in plain English, and explains why the science, not the label, decides these cases now.
The Short Answer
A substantial body of human evidence links glyphosate to non-Hodgkin lymphoma (NHL), though it is not unanimous. In 2015 the International Agency for Research on Cancer (IARC), the World Health Organization’s cancer arm, classified glyphosate as “probably carcinogenic to humans,” naming NHL as the cancer of concern. A 2019 pooled analysis found a 41% higher relative risk of NHL in the most heavily exposed people. The largest U.S. cohort study, by contrast, found no overall association. No study proves glyphosate caused any single person’s cancer — that always depends on the individual’s exposure and diagnosis.
Why This Question Carries More Weight After Durnell
For years, a Roundup plaintiff could argue two things at once: that Monsanto sold a defectively dangerous product, and that it failed to warn consumers of the cancer risk. On June 25, 2026, the U.S. Supreme Court’s decision in Monsanto Co. v. Durnell preempted the failure-to-warn theory where the EPA approved a label with no cancer warning. What survived is the design-defect claim — the argument that the product itself is unreasonably dangerous, independent of anything printed on the label.
That shift moves the science to center stage. A warning claim was partly about what Monsanto knew and hid. A design-defect claim is about the product: is glyphosate, as formulated in Roundup, capable of causing lymphoma? Answering that means engaging the epidemiology directly. So the studies below are no longer background color. They are the case.
What IARC Actually Concluded in 2015
In March 2015, IARC published Monograph Volume 112 and placed glyphosate in Group 2A: probably carcinogenic to humans. That category is worth understanding precisely, because both sides misuse it. Group 2A does not mean “proven to cause cancer,” and it does not mean “a maybe not worth taking seriously.” It means IARC found limited evidence of cancer in humans, sufficient evidence in experimental animals, and — importantly — strong mechanistic evidence that glyphosate can damage DNA and cause oxidative stress. The human cancer IARC pointed to was non-Hodgkin lymphoma.
IARC assesses hazard (can this substance cause cancer under some conditions?) rather than risk at a given dose, which is why regulators using a risk framework can reach different bottom lines without anyone lying. The disagreement between IARC and the EPA is real, and we unpack it separately in IARC vs. EPA on glyphosate. For present purposes, the point is narrower: the world’s leading cancer-classification body reviewed the same literature and concluded glyphosate probably causes NHL.
The Human Studies, One at a Time
Three lines of human evidence do most of the work in this debate. Read together, they explain both the strength of the plaintiffs’ scientific case and the honest uncertainty that remains.
1. The Zhang meta-analysis (2019): a 41% higher risk at high exposure
A 2019 meta-analysis led by Luoping Zhang of the University of California, Berkeley — with co-authors from the University of Washington and the Icahn School of Medicine at Mount Sinai — pooled the available epidemiological studies and focused on the most heavily exposed people in each. It reported a meta-relative risk of 1.41 (95% confidence interval 1.13–1.75) for NHL in the highest glyphosate-exposure groups. In plain terms: the heaviest users had roughly a 41% higher relative risk of developing NHL than unexposed people, and because the confidence interval sits entirely above 1.0, the result was statistically significant. This is the single most-cited number on the plaintiffs’ side, and it comes from mainstream academic researchers, not litigation consultants.
2. The North American Pooled Project (2019): the DLBCL signal
The North American Pooled Project (NAPP) combined two large case-control studies — 1,690 NHL cases and 5,131 controls. Overall, ever having used glyphosate showed a modest, non-significant increase (odds ratio 1.13, 95% CI 0.84–1.51). But when researchers looked at people who handled glyphosate more than two days per year, the risk of NHL nearly doubled (OR 1.73, 95% CI 1.02–2.94), and for one specific subtype — diffuse large B-cell lymphoma, the most common NHL — it more than doubled (OR 2.14, 95% CI 1.07–4.28). The pattern matters: risk rising with the intensity of exposure is a classic dose-response signal, one of the features scientists look for when deciding whether an association reflects causation.
3. The Agricultural Health Study (2018): the null result
Now the evidence that cuts the other way, stated fairly. The Agricultural Health Study (AHS) is a large prospective cohort of more than 54,000 licensed pesticide applicators, published in an updated analysis in the Journal of the National Cancer Institute in 2018. It found no statistically significant association between glyphosate and NHL overall — the rate ratio in the top exposure quartile was 0.87 (95% CI 0.64–1.20). Bayer leans heavily on this study, and it deserves to be taken seriously: it is large, prospective, and enrolled people before they were diagnosed, which avoids some biases that trouble case-control work. The same analysis did note a suggestion of increased acute myeloid leukemia at the highest exposures (rate ratio 2.44), but that finding was not statistically significant and the authors said it needed confirmation.
Why Two Good Studies Point Different Directions
The instinct is to assume one study must be wrong. Usually neither is; they are measuring exposure in different ways, and that difference explains much of the gap.
- Exposure misclassification. The AHS relied on applicators’ own reports of pesticide use over time. If many “unexposed” or low-exposure people were actually exposed, that dilutes a real effect toward the null — making a true association look like no association.
- Cancer latency. Lymphomas can take many years to develop. A cohort followed for a limited window may not capture cancers that surface later, which again biases toward finding nothing.
- The healthy-worker effect. Working applicators tend to be healthier than the general population, which can mask elevated cancer rates in occupational cohorts.
- Study design trade-offs. Case-control studies (like NAPP) can probe intensity and duration of use more finely and detected the strongest signals in heavy users — but are more vulnerable to recall bias. Cohorts (like AHS) reduce recall bias but can wash out real effects. Good scientists weigh these designs differently, which is why expert epidemiologists read the same papers and still disagree.
This is why the debate is not “science vs. junk science.” It is a legitimate methodological argument, and the weight of it — the meta-analysis, the dose-response in heavy users, the subtype-specific DLBCL signal, and the mechanistic evidence — is what led IARC to its Group 2A call.
The 2026 Evidence Update
The literature has not stood still. A systematic review and meta-analysis of glyphosate-based herbicide exposure and NHL, with database searches carried through February 2026, was posted this year on the preprint server medRxiv, continuing the line of subtype-level analysis begun by earlier pooled work. As a preprint it has not yet completed peer review, so it should be read as an update to watch rather than a settled conclusion — but it reflects that researchers are still actively pooling the human data and examining NHL subtypes rather than treating the question as closed. Separately, a 2026 ecological report drew attention for finding that most U.S. counties in the top fifth for glyphosate application had above-average NHL rates; we explain why that county-map pattern is real but cannot prove an individual case in Glyphosate “cancer hot spots”.
The Mechanism: How Glyphosate Could Cause Cancer
Epidemiology tells you whether exposed people get more cancer. Mechanism tells you how a substance could do it — and a plausible mechanism strengthens the inference that a statistical association is causal rather than coincidental. IARC found strong evidence of genotoxicity (glyphosate can damage DNA and chromosomes) and of oxidative stress (an imbalance in which reactive molecules injure cells). Both are recognized routes toward the chromosomal damage seen in lymphomas. Later work has continued in this vein; a 2025 analysis in the AHS itself reported an association between glyphosate use and mosaic loss of the Y chromosome in applicators — a marker of the kind of genetic instability that can accompany blood cancers. The mechanistic picture is part of why the human signal is taken seriously even where the epidemiology is mixed.
What This Means for a Case
Here is the part that trips people up: a civil case does not require the science to be settled or unanimous. Courts decide causation by a preponderance of the evidence — more likely than not — through qualified expert testimony that survives the court’s reliability screening. Juries in state courts have already heard this evidence, weighed it against Bayer’s experts, and found it persuasive enough to support liability on design-defect and negligence theories. (Those are reported outcomes in specific past cases; every case is different, and a past verdict is never a prediction or promise about anyone else’s claim.)
This is where our approach matters. Herb Borroto, M.D., J.D., our Medical-Legal Expert, reads the pathology in each case personally — the biopsy, the immunophenotyping that identifies the specific NHL subtype, and the staging — because the causation argument is far stronger when the diagnosis is pinned down precisely and matched to the exposure history and the subtype-level science. Alex Alvarez, our Managing Partner and a Board Certified Civil Trial Lawyer, has built these cases on the design-defect and strict-product-liability theory from the start — the theory Durnell left standing, and the one that turns on exactly the evidence surveyed here.
The threshold question for any individual is simpler than the scientific debate: a qualifying diagnosis — non-Hodgkin lymphoma and its subtypes, chronic lymphocytic leukemia, or multiple myeloma — connected to a documented history of Roundup exposure. Our overview of who qualifies for a Roundup lawsuit walks through how that evaluation works.
Frequently Asked Questions
Does glyphosate cause non-Hodgkin lymphoma?
The scientific evidence is genuinely divided, but a substantial body of it links glyphosate to non-Hodgkin lymphoma. In 2015 the World Health Organization’s cancer agency (IARC) classified glyphosate as probably carcinogenic to humans, citing NHL as the cancer of concern. A 2019 meta-analysis led by Luoping Zhang reported that people in the highest glyphosate-exposure groups had a 41 percent higher relative risk of NHL. The Agricultural Health Study, a large U.S. cohort, found no overall association. No study proves that glyphosate caused any one person’s cancer; that always depends on the individual’s exposure history and diagnosis.
Why do different glyphosate studies reach different conclusions?
The two most-cited studies measure exposure differently and that drives much of the disagreement. Case-control studies like those pooled in the North American Pooled Project ask people about their herbicide use and found roughly a two-fold increased risk of diffuse large B-cell lymphoma among the heaviest users. The Agricultural Health Study, a prospective cohort, enrolled applicators before diagnosis and found no overall NHL association. Cohorts reduce recall bias but can dilute a real effect through exposure misclassification and long cancer latency. Scientists weigh these designs differently, which is why reasonable experts read the same data and disagree.
How could glyphosate cause cancer at the cellular level?
IARC found strong evidence that glyphosate is genotoxic — capable of damaging DNA — and that it induces oxidative stress, a state where reactive molecules injure cells. Both are recognized pathways toward the kind of chromosomal damage seen in lymphomas. This mechanistic evidence is part of why IARC reached a Group 2A classification even where the human studies were mixed: a plausible biological mechanism strengthens the case that an observed statistical association reflects real causation rather than coincidence.
Does the science have to be settled for a Roundup lawsuit to move forward?
No. Civil cases are decided by a preponderance of the evidence — more likely than not — not by scientific certainty or unanimous consensus. A plaintiff proves causation through qualified expert testimony that survives the court’s reliability screening, drawing on the epidemiology, the mechanistic evidence, and the individual’s medical records. After the Supreme Court’s Durnell ruling preempted the failure-to-warn theory, the design-defect claim carries more of the case, and that claim turns directly on this body of scientific evidence about the product itself.
Bottom Line
“Does glyphosate cause lymphoma?” has no one-word answer, and anyone who gives you one is selling something. What the human evidence shows is a real, well-documented association — strongest among the heaviest users and for the DLBCL subtype — supported by a plausible biological mechanism and endorsed by the World Health Organization’s cancer agency, alongside one large cohort study that found no overall effect. That is enough of a scientific foundation for courts to let qualified experts testify and juries to decide. And after Durnell pushed the case onto the design-defect theory, this evidence is not a sidebar — it is the heart of the matter.
Nothing here is a prediction about any particular case, and past verdicts are never a promise of any future result. This article is general legal and scientific information, not advice on your own facts. If you or a family member were diagnosed with Non-Hodgkin Lymphoma, B-cell Lymphoma, Chronic Lymphocytic Leukemia, or Multiple Myeloma after Roundup exposure, a free case review is the right next step — no obligation, and no fee unless we recover for you.