The claim that taking acetaminophen (Tylenol) during pregnancy causes autism in children is not supported by the strongest available evidence
Where the evidence lands: ContradictedThat acetaminophen (Tylenol) taken by a pregnant woman crosses to the fetus and causes or substantially raises the risk of autism, so that the rise in autism diagnoses can be traced in significant part to widespread use of the drug in pregnancy, and that women should therefore avoid it.
Believed by: A causal link is rejected by ACOG, the American Academy of Pediatrics, the Society for Maternal-Fetal Medicine, and, after review, the World Health Organization. The claim gained wide public traction after the September 2025 federal announcement, and remains contested in politics even as the strongest studies and major medical bodies find no established causal effect.
The full story
What is actually claimed, and what is not
Two very different statements get blurred together in this story, and keeping them apart is the whole task. The first is narrow and defensible: some studies have reported a statistical association between acetaminophen use in pregnancy and later diagnoses of autism or ADHD. That is true, and this file does not deny it. The second is broad and unsupported: the drug causes autism, so that widespread use in pregnancy is driving the rise in diagnoses and women should avoid it. The rated claim, and the thing the strongest evidence contradicts, is the second one.
Acetaminophen, sold in the United States as Tylenol and known internationally as paracetamol, is one of the very few pain and fever relievers considered appropriate in pregnancy, which is exactly why the question matters to so many people. It is used by a large share of pregnant women, often for fever or infection, conditions that themselves carry risk if left untreated. That ubiquity is part of why a scare around it spreads so far, and part of why getting the reasoning right is worth the effort.
So the question is not whether an association has ever been reported. It has. The question is whether that association reflects the drug itself, or the reasons women take it, and what happens to the signal when a study is designed to tell those apart.
Correlation, causation, and the confounding trap
Observational studies can only watch what people already do; they cannot randomly assign a drug. That makes them vulnerable to confounding: a third factor that drives both the exposure and the outcome and fakes a link between them. Here the confounder is not exotic. Women who take acetaminophen in pregnancy differ systematically from those who do not. They have more fevers, more infections, more pain, and more chronic conditions, and some of those conditions carry their own genetic loading for neurodevelopmental differences.
This is called confounding by indication: the reason for taking the drug, not the drug, may be doing the work. A mother with more genetic liability for autism or ADHD tends to report more pain and to use more acetaminophen, and can pass that liability to her child through genes, not through a pill. Untreated fever and infection in pregnancy are themselves associated with worse outcomes. Every one of these pathways can manufacture a correlation between the drug and a later diagnosis while the drug does nothing at all.
Measuring the exposure more precisely does not fix this. The much-cited 2019 study measured acetaminophen biomarkers directly in umbilical-cord blood, a real methodological strength, but a sharper exposure measurement in an observational design still cannot separate the drug from the reasons it was taken. To break the confounding, you need a design that holds the family factors constant.
The question was never whether a correlation exists. It was whether the correlation survives once you account for why the drug was taken.
The study built to break the confounding
That design arrived in 2024, when JAMA published a Swedish nationwide study of roughly 2.4 million children born between 1995 and 2019, led by researchers at the Karolinska Institute and Drexel University and funded by the U.S. National Institutes of Health. Its key move was a sibling-control analysis: comparing siblings from the same family who had different acetaminophen exposure in the womb. Siblings share much of their genetics and their home environment, so if the drug were the cause, the exposed sibling should show higher risk than the unexposed one.
In conventional models, the study reproduced the familiar small association. Then, in the sibling comparisons, it disappeared. The hazard ratio for autism was 0.98 (95% confidence interval 0.94 to 1.02), essentially no effect, with similarly null results for ADHD and intellectual disability. The authors concluded that the earlier signals were best explained by familial confounding rather than by any causal effect of the drug. A 2025 review in the International Journal of Gynecology and Obstetrics reached the same bottom line: the evidence does not support a causal association.
No single study is the last word, and observational data are always limited at the extremes of very heavy or prolonged use. But this is the largest and best-designed test to date, and it points the same way the careful reading of the earlier work already did: when you remove the shared family factors, the link is not there.
The 2025 announcement, reported as a dispute
On 22 September 2025, at a White House event, President Donald Trump and HHS Secretary Robert F. Kennedy Jr.announced a set of actions on autism, among them a warning advising against acetaminophen use in pregnancy. The FDA issued a notice to physicians and began the process toward a possible label change, and Trump urged pregnant women to “tough it out.” That is what was said, and it is reported here plainly and by attribution, because it is a real and consequential event.
What it was not is a new scientific finding. The response from the medical world was immediate and broad. ACOG reaffirmed acetaminophen as the analgesic and antipyretic of choice in pregnancy and said the current weight of evidence does not support a causal link. The American Academy of Pediatrics, the Society for Maternal-Fetal Medicine, and, after review, the World Health Organizationsaid much the same. FactCheck.org and other outlets catalogued the problems with the studies invoked. The drug's maker, Kenvue, formally asked the FDA to reject the labeling petition as unsupported by the science, and within weeks the federal messaging softened toward describing a possible association rather than a proven cause.
The honest way to hold this is as a dispute with a factual center. A government may issue a warning; that is a policy choice, and it is recorded here as one. It does not, on its own, establish the causal claim, and the bodies that synthesize the evidence rejected it. This file makes the announcement visible without adopting its conclusion.
A warning from a podium is a policy act, not a peer review. The weight of evidence is what the medical bodies read across many studies, and here it pointed the other way.
Where this leaves a reader
Keep the layers apart. There is a documented statistical association in some observational studies, and it is fair to say so. There is an unsupported causal claim built on top of it, and the strongest study to date, the 2024 sibling-controlled analysis of about 2.4 million children, found the association evaporates once family factors are held constant. And there is a 2025 policy dispute, in which a federal warning collided with the near-uniform position of the major medical bodies. Those are three different things, and the rating here attaches only to the second: the causal claim is not supported by the strongest evidence.
That is also why the rise in autism diagnoses does not need a single pill to explain it. Researchers attribute it largely to broadened diagnostic criteria, more screening and awareness, and a substantial genetic component. A one-cause story is emotionally tidy, but it erases that picture, and it can do harm: the clearest risk in the whole episode, ACOG and others warn, is a frightened patient leaving a high fever or severe pain untreated, which carries documented dangers of its own.
One last point, and it is the important one. This page exists to explain how the evidence is weighed, not to tell anyone what to take. Decisions about pain and fever in pregnancy are medical decisions that depend on the individual, and they belong with a clinician who knows the patient. Read the reasoning here, then follow your doctor's guidance, not a website and not a viral clip.
Watch
What's still unexplained
- Why do raw associations keep appearing at all? The best current answer is confounding by indication (the reasons for taking the drug, from fever to genetics, track with neurodevelopmental outcomes), which is precisely why sibling-controlled designs that hold family factors constant are treated as the more trustworthy test.
- Could very heavy or prolonged use, as opposed to ordinary short-term use, carry any effect? The strongest studies do not establish one, but observational data are always limited at the extremes, and researchers continue to examine dose and duration rather than treating the question as fully closed.
- How did a contested observational signal become a national warning? The 2025 episode is as much about how science is communicated and politicized as about the drug itself, and the gap between the podium and the peer-reviewed literature is part of what this case documents.
- What actually drives the rise in autism diagnoses? The consensus points to expanded diagnostic criteria, increased screening and awareness, and a large, mostly genetic component, not to any single environmental trigger, which is the broader context a one-cause story tends to erase.
Point by point
The claim: Multiple studies found that children exposed to acetaminophen in the womb were more likely to be diagnosed with autism.
What the record shows: Several observational studies did report such associations, and that part is not in dispute. The key point is what an association can and cannot show. These studies compare unrelated children and cannot fully separate the drug from the reasons it was taken. An association is a starting point for investigation, not proof of cause, and the strongest later work was designed specifically to test whether this particular association held up.
The claim: The 2019 cord-blood study measured the drug itself in the baby's blood, so it must show a real biological effect.
What the record shows: Measuring acetaminophen biomarkers in cord blood makes the exposure more precise, which is a genuine strength, but it does nothing to fix confounding: a mother's fever, infection, pain, or underlying genetics can both lead to acetaminophen use and independently affect neurodevelopment. A more accurate exposure measurement in an observational design still cannot establish that the drug, rather than the reason for taking it, is responsible.
The claim: The association is the same thing as causation, so avoiding the drug will lower autism risk.
What the record shows: This is the central error the strongest study was built to catch. In the 2024 JAMA analysis of about 2.4 million Swedish children, a small association appeared in ordinary models but vanished in sibling comparisons, where one sibling was exposed and another was not within the same family, holding shared genetics and home environment constant. The autism hazard ratio was 0.98, with ADHD and intellectual disability similarly null. That pattern is the signature of confounding, not of a causal drug effect.
The claim: The reasons a woman takes acetaminophen have nothing to do with her child's neurodevelopment.
What the record shows: The evidence points the other way. Mothers with more genetic liability for autism or ADHD tend to report more pain and are more likely to use acetaminophen, and untreated fever and infection in pregnancy carry their own documented risks. FactCheck.org and researchers note additional problems in some earlier studies, including non-random recruitment of participants already seeking answers about autism. These are exactly the pathways that make a raw correlation misleading.
The claim: Medical authorities are ignoring the risk and telling women the drug is perfectly safe.
What the record shows: That misstates the guidance. ACOG reaffirms acetaminophen as the preferred option for pain and fever in pregnancy while advising, as it long has, judicious use at the lowest effective dose for the shortest necessary time, in consultation with a clinician. Recommending measured use is not the same as denying a risk; it is the standard framing for any medication in pregnancy, and it reflects a body of evidence that has not established a causal harm.
The claim: The September 2025 federal warning shows the government has concluded the drug causes autism.
What the record shows: The announcement was a policy and communications action, not a new scientific finding, and it drew immediate criticism from ACOG, the American Academy of Pediatrics, the Society for Maternal-Fetal Medicine, the WHO, and the manufacturer. Within weeks the messaging shifted toward describing a possible association rather than a proven cause. A government statement is reported here as an attributed position in a live dispute; it does not convert an unproven claim into an established fact.
The claim: If pregnant women simply avoid the drug to be safe, there is no downside.
What the record shows: Clinicians warn there is a real downside. The main risk of the fear message, ACOG and others note, is that pregnant patients may leave a high fever or severe pain untreated, and untreated fever and pain in pregnancy can themselves harm both mother and fetus. 'Just avoid it' is not a neutral default; it trades a hypothetical, unestablished risk for documented ones, which is why the guidance is to decide with a clinician rather than to abstain out of alarm.
Other readings
Angles that don't fit neatly into the claim or its rebuttal, laid out and weighed, not endorsed.
The precautionary-statement reading
Some scientists, including the authors of the 2021 consensus statement, argued that even uncertain evidence warranted caution and lower use as a hedge. Taken narrowly, 'use the lowest effective dose for the shortest time' is unobjectionable and in fact matches ACOG's standard advice for any drug in pregnancy. The problem is the slide from that measured position to a flat causal claim and a 'don't take it' message: the precautionary framing does not establish that the drug causes autism, and treating it as if it does inverts the evidence.
Why 'follow the guidance, not the headline' is the real takeaway
The durable lesson here is about how to read a health scare. A single study, a biomarker, or a government podium is not the weight of evidence; the weight of evidence is what bodies like ACOG synthesize across many studies, giving extra credit to designs that can rule out confounding. This file exists to explain that reasoning, not to issue a medical instruction. Decisions about pain and fever in pregnancy should be made with a clinician who knows the patient, not from a website or a viral clip.
Timeline
- 2019-10A JAMA Psychiatry study of the Boston Birth Cohort reports that biomarkers of acetaminophen measured in umbilical-cord blood are associated with higher odds of later ADHD and autism diagnoses. It is widely covered and becomes a frequently cited anchor for the concern, though it is observational and cannot establish cause.
- 2021-09A group of scientists and clinicians publishes a consensus statement in Nature Reviews Endocrinology urging precautionary limits on acetaminophen use in pregnancy, citing accumulated observational associations with neurodevelopmental outcomes. Other researchers respond that the evidence does not justify alarming pregnant patients away from one of the few pain relievers considered safe for them.
- 2023Reviews and editorials increasingly stress the central methodological problem: women who use acetaminophen in pregnancy differ systematically from those who do not (more fevers, infections, pain, and chronic illness), so a raw association may reflect the reasons for use, a phenomenon called confounding by indication, rather than any effect of the drug.
- 2024-04JAMA publishes a Swedish nationwide study of about 2.48 million children born 1995-2019 (Ahlqvist and colleagues, Karolinska Institute and Drexel University, NIH-funded). In conventional models a small association appears, but in sibling-control analyses, comparing siblings differently exposed within the same family, the association disappears (autism hazard ratio 0.98). The authors conclude the earlier signals likely reflect familial confounding, not causation.
- 2025-08The FDA's own acetaminophen information page states the agency has not found clear evidence that appropriate use of the drug in pregnancy causes adverse developmental outcomes, reflecting the mainstream regulatory position going into the autumn.
- 2025-09-22At a White House event, President Trump and HHS Secretary Robert F. Kennedy Jr. announce actions on autism, including a warning advising against acetaminophen use in pregnancy; the FDA issues a physician notice and begins the process toward a label change. Trump urges pregnant women to 'tough it out.' ACOG, the maker Kenvue, and other medical bodies push back the same day.
- 2025-09ACOG issues a practice advisory and a news release affirming acetaminophen as the analgesic and antipyretic of choice in pregnancy and stating that the current weight of evidence does not support a causal link to neurodevelopmental disorders. The World Health Organization and the American Academy of Pediatrics issue statements to similar effect.
- 2025-10The proposed federal messaging is walked back in tone toward describing a possible association rather than a proven cause, and the professional-label proposal frames the relationship as suggestive rather than established. Kenvue formally asks the FDA to reject the labeling petition, arguing the change is unsupported by the science.
Contradicted. The specific claim rated here is causal: that acetaminophen taken in pregnancy causes autism. That claim is not supported by the strongest evidence, and the leading medical bodies reject it. Some observational studies did report a statistical association between prenatal acetaminophen use and later autism or ADHD diagnoses, and that association is real in the data; the dispute is over what it means. The best-designed study to date, a 2024 JAMA analysis of about 2.4 million Swedish children that used sibling controls to strip out shared family genetics and environment, found the association vanished once those confounders were removed (autism hazard ratio 0.98). The American College of Obstetricians and Gynecologists (ACOG) reaffirms acetaminophen as the pain and fever reliever of choice in pregnancy and says the weight of evidence does not support a causal link. In September 2025 the Trump administration and HHS Secretary Robert F. Kennedy Jr. announced a warning tying the drug to autism; that policy step, and the medical pushback against it, are reported here as an attributed dispute, not as evidence that the causal claim is true. This file rates the causal claim debunked; it does not tell readers what to take. Follow your clinician's guidance, not a website.
Reviewed by The Conspiratory Editors · Last reviewed July 19, 2026 · How we rate
Sources
- 1.Acetaminophen Use During Pregnancy and Children's Risk of Autism, ADHD, and Intellectual Disability, JAMA (Ahlqvist et al.) (2024)
- 2.Study reveals no causal link between neurodevelopmental disorders and acetaminophen exposure before birth, National Institutes of Health (2024)
- 3.ACOG Affirms Safety and Benefits of Acetaminophen during Pregnancy, American College of Obstetricians and Gynecologists (2025)
- 4.Acetaminophen Use in Pregnancy and Neurodevelopmental Outcomes, ACOG (Practice Advisory) (2025)
- 5.President Trump, Secretary Kennedy Announce Bold Actions to Tackle Autism Epidemic, U.S. Department of Health and Human Services (2025)
- 6.Trump Administration's Problematic Claims on Tylenol and Autism, FactCheck.org (2025)
- 7.Trump blames Tylenol for autism. Science doesn't back him up, NPR (2025)
- 8.Tylenol's maker pushes back against possible label change linking pain reliever's use in pregnancy to autism, CNN (2025)
- 9.Acetaminophen is Safe for Children When Taken as Directed, No Link to Autism, American Academy of Pediatrics (2025)
- 10.WHO statement on autism-related issues, World Health Organization (2025)
Help us investigate
This is a living case file. If you spot an error or know evidence we missed, tell us, and weigh in on where you land.
Where do you land?
Cast your read on this one.
Comments
Add your take. Comments are read and approved by a human before they appear, so keep it on topic and civil. Please do not accuse named, living people of crimes.