Is Roundup Dangerous to Humans?

Where the Disagreement Comes From

In 2015 the International Agency for Research on Cancer, part of the World Health Organization, put glyphosate in its Group 2A: probably cancer-causing in people. The US Environmental Protection Agency reviewed its own evidence and reached the opposite view, that it is not likely to cause cancer at the doses ordinary use involves.

Those two answers get quoted against each other constantly. What usually gets left out is what happened next.

What Happened to the EPA Finding

In June 2022 a federal appeals court threw that EPA conclusion out. The Ninth Circuit held it was not backed by enough evidence, and said the agency had reasoned inconsistently, had not followed its own guidelines for judging cancer risk, had passed over studies, and had set aside advice from its own panel of scientists. The court also noted that the EPA's own analysis held signs of a link to non-Hodgkin lymphoma that the agency never properly accounted for.

The court did not rule that glyphosate causes cancer. It said the EPA might well land in the same place again, but would have to show its work. A few months later the EPA withdrew the decision altogether.

Four years on it has still not issued a replacement. As of August 2026 the agency was publishing a search of the literature to feed into a new assessment. Its position is that the underlying science it relied on has not changed. But as things stand, the American regulator has no cancer finding in force, which is worth knowing whenever someone tells you the EPA has cleared it.

What "Probably Carcinogenic" Actually Means

The label describes how strong the evidence is that something can cause cancer. It does not say how much your risk goes up, or at what dose. The same group includes things most people would not think twice about. So the classification means something, but it is not a finding that using Roundup will give you cancer.

What the Best Studies Found

The strongest single study is the Agricultural Health Study, which has followed around 54,000 licensed applicators over many years. Its 2018 update found no link between glyphosate and non-Hodgkin lymphoma, or its subtypes, or solid tumours. It did see a possible rise in a different blood cancer among the most exposed, which the authors said needed confirming.

That study is the best evidence the companies have, and any honest page has to say so.

The other side leans on a 2019 analysis that pooled that same study with five others and looked only at the most heavily exposed people in each. It found their risk of non-Hodgkin lymphoma was 41% higher.

Why the Same Data Gives Two Answers

Both of those are real findings from overlapping research. What reconciles them is a later analysis that ran the numbers several ways.

Count everyone who ever used glyphosate, and assume no delay between exposure and illness, and the pooled result is no measurable difference at all. Count only the heaviest users, and assume the cancer takes twenty years to show up, and you get the 41% figure. Assume fifteen years instead and it drops to about 25%.

Same studies. Opposite headlines. The whole gap sits in two choices: who counts as exposed, and how long you think the illness takes to appear.

What the Science Still Cannot Tell You

Knowing where the weak points are is more useful than picking a side.

• Nobody can measure past exposure. Almost all of this rests on people remembering how much they sprayed decades ago. There is no blood test that reconstructs it.
• The delay is guesswork. This kind of cancer may take fifteen to twenty years to appear. Earlier follow-up periods were far shorter, which may be why they saw nothing.
• Farmers use many chemicals. Separating one out from everything else sprayed over a career is genuinely hard.
• The tests are on the wrong thing. Most lab work uses pure glyphosate. People are exposed to mixed products containing other ingredients that help it stick and soak in, and those are studied far less.
• It is not one illness. Non-Hodgkin lymphoma covers many subtypes that may have different causes, and studies often pool them.

Why Dose and Duration Matter

With most chemicals, risk depends on how much and how long. The people in these lawsuits are mostly not weekend gardeners. They are farm workers, landscapers and grounds crews who handled the product regularly for years. That difference shapes how a case is looked at.

Linked To Is Not Caused By

When researchers find a link, they mean an illness shows up more often in a more exposed group. That is not proof the chemical caused any one person's cancer. In court those are two separate questions, and a plaintiff has to win both: that the product can cause this illness, and that it caused theirs.

What Juries Have Decided

Juries have heard this evidence dozens of times and have not agreed either. Bayer, which bought Monsanto in 2018, says it has won 17 of its last 25 trials. The trials it lost produced some of the largest verdicts in American history: about $2.05 billion in California, $2.25 billion in Philadelphia in January 2024, and roughly $2.1 billion in Georgia in March 2025. A $1.56 billion Missouri verdict was cut to $611 million and upheld on appeal.

Look closely at those numbers and something shows through. In the Georgia case, $65 million was to compensate the man who got sick and $2 billion was to punish the company. In Philadelphia it was $250 million against $2 billion. Punishment ran twenty to thirty times the compensation, over and over.

Punitive damages are not compensation. They are what a jury awards when it decides a company behaved badly enough to be punished. So the size of those awards is a verdict on conduct, and to understand them you have to look at what the jurors were shown.

What the Jurors Were Shown

Suing the company meant it had to hand over its internal records. Millions of pages came out, and lawyers have put the same handful of documents in front of jury after jury. They became known as the Monsanto Papers. Four things in them did most of the damage.

Company staff discussing putting outside names on their own work. In a 2015 email a Monsanto scientist floated having the company write scientific material and having outside experts, in his words, “just edit & sign their names so to speak.” He pointed back to a review published in 2000 as the model. Regulators had treated that 2000 review as independent. A separate company memo from the same year used the word ghostwrote about another cancer paper. Monsanto denied ghostwriting the 2000 review and said the accusation rested on one email pulled out of ten million pages.

A friendly contact inside the EPA. Jurors saw an internal email describing a senior EPA official, who chaired the committee assessing glyphosate's cancer risk, saying he deserved a medal if he could stop a review of the chemical by a different federal health agency. Another company email told colleagues not to get their hopes up about it. The official had retired and was never a defendant, and nothing was proven against him. But the exchange landed badly with juries, because the company's whole defense rests on the EPA having cleared the product.

An expert who was not what the company wanted. In the late 1990s Monsanto brought in an outside specialist on genetic damage. What he reported back was unwelcome enough that an internal email described the company as being in a hole over it, and a colleague suggested dropping him. Jurors were left to draw their own conclusion about what happens to inconvenient findings.

Money spent on the response rather than the question. Testimony in one case put the company's budget for countering the 2015 cancer classification at roughly sixteen to seventeen million dollars in a single year. Set against a company that had not run a long-term cancer study on the mixed product people actually spray, that contrast is easy for a jury to hold on to.

That last point connects back to the gaps above. One of the weakest spots in the science is that the testing was done on the active ingredient rather than the products sold. In court, plaintiffs argued that gap was not an accident.

None of this is evidence that glyphosate causes cancer. It is evidence about how a company behaved, and that is a different question. It is also why compensatory awards in these cases have often been modest while the punishment figures ran into the billions.

The Verdict That Explains the Split

The most revealing outcome came in Philadelphia in September 2024. The jury decided Roundup was defective, and decided it had not caused that particular man's cancer. Both at once. That is the two-question split above, worked through by twelve people in a real case, and it is why no verdict anywhere settles what happened to any other person.

So Where Does That Leave You

Nowhere tidy, which is the truthful answer. There is real evidence of a link at heavy, long-term exposure, real evidence against it, no way to measure what anyone was actually exposed to years ago, no regulator finding currently in force in the US, and juries that keep splitting.

For reported effects, see Glyphosate Exposure Symptoms. For the illness itself, see Roundup and Non-Hodgkin Lymphoma. For the legal picture, see Roundup Cancer Lawsuits, Roundup Lawsuit Updates, and Glyphosate Lawsuits: EPA, Bayer and the Supreme Court.