The Tylenol Autism Studies, Explained

The Early Studies

Over the past decade several studies followed large groups of children from birth. A number of them found that children whose mothers reported using more acetaminophen during pregnancy were diagnosed with autism or ADHD somewhat more often.

Some of those drew a lot of attention and prompted calls for caution. They are useful for spotting a signal. Their weakness is that they compare groups of people who differ in all sorts of ways besides the one being studied.

The Confounding Problem

Confounding is the word for everything that travels alongside the thing you are measuring, and here it is the whole difficulty.

Nobody takes acetaminophen for no reason. They take it for a fever, an infection, pain, or inflammation, and any of those might matter to how a pregnancy develops. Autism also has a strong genetic component, and parents who reach for medication more often may differ from those who do not in ways that get passed on. A study that cannot separate the drug from all of that may be measuring the reason rather than the pill.

The 2024 Sibling Study

In 2024 a large study tackled that directly. Published in JAMA, it compared brothers and sisters within the same families, where the mother had used acetaminophen in one pregnancy and not another.

Two children of the same mother share her genes, her health and the household they grew up in. So if one pregnancy involved acetaminophen, the other did not, and both children turn out the same, none of the things they have in common can be the explanation. In that analysis the link disappeared. It is the single most influential result in this area and a major reason scientific bodies have held back from saying acetaminophen causes autism.

The Two Studies the Government Pointed To

When the FDA acted in September 2025 it named two pieces of research, and they are not the weak kind. Each was built to get around the confounding problem, by a different route.

One, published in the American Journal of Epidemiology, followed 8,856 children and checked itself in an unusual way. It looked at acetaminophen use about four years before the pregnancy and about four years after, as well as during. Anything fixed about the mother should have shown up in all three windows. Only the pregnancy window showed a link to ADHD.

The other, published in JAMA Psychiatry, did not ask anyone to remember anything. When a baby is born there is blood left in the umbilical cord, carrying whatever crossed over from the mother during the pregnancy, and some hospitals freeze a sample. Researchers tested 996 of those frozen samples for acetaminophen and then looked at how the children were doing roughly ten years later. Those with the most in their blood had markedly higher rates of ADHD and autism than those with the least, rising as the amount rose.

That second one matters for a reason worth flagging: it did not depend on anyone remembering anything, which is the usual complaint about this field. Its limits are elsewhere. The sample is small, it comes from a single city hospital, and the children in it were diagnosed at rates far above the general population, so it describes a higher-risk group than most families reading this.

Why the Studies Disagree

Not because anyone is lying. Because of how each was built.

Studies differ in how they worked out who was exposed. Often that comes down to a parent remembering years later, though not always, as the umbilical cord work above shows. They differ in whether they could account for why the drug was taken. And they differ in whether they could control for family factors at all.

A fair reading of the whole body of work is that careful studies land on both sides. The sibling comparison is the largest and removes what families share. The umbilical cord study is the most direct measure of what reached the child. They point different ways, and neither one settles it. That leaves causation unproven and argued over. In July 2026 a federal appeals court said something parallel in the litigation, holding that the plaintiffs' experts had offered a permissible reading of evidence that scientists genuinely disagree about. That ruling was about what a jury may hear, not about which reading is right.

The 2025 White House Announcement

On September 22, 2025 the President, the Health Secretary and the FDA Commissioner held a press conference on autism, and the FDA started the process of changing the acetaminophen label.

It is usually described as a government report. It was not one. A comprehensive report had been promised for that month and never arrived. The main piece of science behind the announcement was a 2025 review out of Mount Sinai that pulled together 46 earlier studies. Reviews like that gather what already exists rather than producing new evidence, so it did not settle anything the individual studies had left open. The FDA said in the same breath that a causal relationship has not been established.

The report raised the profile of the question and became part of the backdrop to the court cases. It did not answer it. Where the cases stand now is on Tylenol Lawsuit Updates, and the legal overview is on Tylenol Autism Lawsuits.

How to Read the Next Headline

Another study will come along. Three questions will tell you most of what you need.

• What kind of study is it, and did it compare people within families?
• How did it work out who was exposed, and how much?
• Could it account for why the drug was taken, and for genetics?

One study reporting a link is a much weaker basis for a decision than the weight of well-built evidence. Which is the reason that decision belongs with your own doctor. See also Tylenol, Acetaminophen, and ADHD.

Common Questions

What do the Tylenol autism studies show?

They disagree. Several found autism and ADHD turning up more often among children whose mothers used acetaminophen. A large 2024 study comparing siblings found that link largely disappeared. How each was built explains most of the gap.

Why do the studies reach different conclusions?

Study design. Comparing unrelated families makes it hard to separate the drug from the fever it was taken for, or from what runs in a family. Comparing two children of the same mother rules out anything they both inherited or grew up with, and those studies tend to find little or nothing.

Is the research settled?

No. The September 2025 announcement raised attention without answering the question, and the FDA still says causation is not established.