Does Tylenol in Pregnancy Cause Autism or ADHD?

Why the older studies looked scary

The concern isn't imaginary. Meta-analyses pooling standard cohort studies found children whose mothers used acetaminophen in pregnancy had roughly a fifth to a third higher rate of ADHD and about a sixth higher rate of autism, with larger signals for long courses and third-trimester use. Small biomarker studies measuring acetaminophen metabolites in cord blood or meconium reported even higher odds, sometimes two or three times baseline. That's the evidence base the FDA drew on when it sent its September 2025 notice.

The problem is what those studies could not measure. Women who take acetaminophen in pregnancy differ from those who don't in ways that also affect the child: fevers, infections, migraines, chronic pain, and heritable traits like maternal ADHD. Standard cohorts adjust for education, smoking, and a handful of variables they can capture. They can't adjust for the genes and home environment a mother passes on whether or not she took the pill.

What sibling comparisons changed

A sibling design fixes that. If the same mother had one pregnancy on acetaminophen and one off it, everything she brings to the table stays constant, and only the exposure changes. When Swedish researchers ran that comparison across 2.48 million children born from 1995 to 2019, the risk elevations from the whole-population model disappeared. Autism, ADHD, and intellectual disability all landed at essentially the baseline rate in exposed siblings, and higher cumulative doses did not push the risk up.

The pattern has held internationally. The Hong Kong cohort of 708,020 mother-child pairs, published in JAMA Internal Medicine, ran sibling-matched analyses on the subset of families with a discordant sibling and found no difference across any trimester or dose. Norwegian and Japanese sibling analyses land in the same place. Negative-control analyses, comparing children of fathers who used acetaminophen or of mothers who used it before pregnancy, show similarly elevated risks. Those children weren't exposed in the womb, so the elevation points to traits the family shares. The 2025 BMJ umbrella review, synthesizing nine reviews and 40 studies, concluded the earlier signals largely dissolve once familial confounding is accounted for.

The absolute numbers were small even in the alarming version. At age 10, the exposed-versus-unexposed gap in the Swedish population model was about one extra autism diagnosis per thousand children and roughly two extra ADHD diagnoses per thousand. Once siblings were compared, even that closed. Acetaminophen for a real fever or migraine in pregnancy isn't something to feel guilty about.

The alternative to treating a fever isn't 'no exposure'

For most pregnant women, the choice is between treating a fever and leaving it running. An untreated maternal fever isn't benign. First-trimester fevers raise the risk of neural tube defects and congenital heart defects. A fever late in the first trimester is linked to about a third higher rate of ADHD in the child, independent of anything used to treat it. Intrapartum hyperthermia is starker still: at 39°C or higher, the risk of neonatal encephalopathy runs several times baseline.

NSAIDs are not a clean swap. In the second half of pregnancy they can constrict the fetal ductus arteriosus and drop amniotic fluid. Opioids bring their own risks, including neonatal withdrawal and addiction. That's why ACOG in September 2025 and SMFM in June 2026 both reaffirmed acetaminophen as first-line for pain and fever, calling the suggestion of a causal autism link unsupported by the strongest evidence. The FDA's own notice granted that a causal relationship has not been established.

Proposed biological mechanisms include oxidative stress from a reactive metabolite, altered prostaglandin signaling, and endocrine effects. None has been demonstrated in human fetal neurodevelopment, and rodent studies at therapeutic doses have not shown consistent neurotoxicity.

What would change the answer

Two things would change this. A within-family design that showed a real dose-response after full control for maternal illness and genetics would suggest the current sibling studies are missing something. So would a mechanism confirmed in human fetal tissue. A randomized trial in pregnancy would settle it, and one has been proposed, though the ethics are hard.

So the practical answer is straightforward. Treat a fever, a bad headache, or real pain with acetaminophen at the lowest dose that works, for as short a time as you need. Talk to your OB before long or repeated courses, both to check what's driving the symptoms and to weigh alternatives that fit your situation.