Science Spotlight:
Study Links Grandfathers’ Smoking and Drinking to Autism Risk in Grandchildren

Barbara Cohn, Ph.D., has spent her career following one extraordinary group of families. Between 1959 and 1967, thousands of pregnant women in California agreed to participate in the Child Health and Development Studies (CHDS), a longitudinal cohort following more than 15,000 families and three generations of births in California.  In interviews, they described their diets and daily lives, and how much they and their husbands smoked and drank. They answered plainly, because at the time neither smoking nor drinking was widely understood to be a risk in pregnancy. Six decades later, those candid answers have become one of the few datasets in the world capable of asking a question most studies cannot: does an exposure in one generation reach into the third?

The answer, in a new study published in Autism Research, appears to be yes, and the signal comes not from the grandmothers, but from the grandfathers.

The Finding

The analysis covered 6,336 grandparents who had at least one of the exposures studied: tobacco, alcohol, coffee and tea. The grandchildren of men who drank five or more servings of hard liquor a week during their wife’s pregnancy were nearly three times as likely to be diagnosed with autism spectrum disorder as the grandchildren of non-drinkers. Tobacco showed a similar pattern: any use—pipes, cigars, cigarettes—raised the risk, and grandchildren of men who smoked 20 or more cigarettes a day carried roughly double the risk. Accounting for age, education and race did not eliminate the associations. The associations are statistically significant, but they are not proof of cause.

Heavy drinking by the father’s mother was linked to higher autism risk too, but so few grandmothers drank heavily that the researchers could not be confident in that result. Coffee or tea consumption by either grandparent showed no comparable association.

There is also a historical dimension to when these exposures happened. Smoking and heavy drinking were near-universal among American men in the 1950s and 1960s.

Why researchers are looking at the germline

Autism has both genetic and environmental contributors, including prenatal nutrition and chemical exposures. Prevalence has been rising worldwide, a trend attributed largely to broader diagnostic criteria and better detection. One possible explanation for a three-generation signal involves the germline epigenome, the chemical tags that help determine which genes are switched on. Those tags may be disturbed by chemical exposures, and such disturbances in the cells that become sperm could persist in the next generation’s germline. The exact mechanism is not yet known, but the quality of the data in this study suggests that a man’s exposures around the time of conception might alter reproductive cells in a way that contributes to autism risk two generations later.

Why these answers are unusually reliable

This is what scientists call a prospective study: the habits were recorded first, and the health outcomes came later. That design guards against recall bias, the tendency of people who already know an outcome to remember their habits differently. The grandmothers reported their smoking and drinking and that of their spouse in the early 1960s, before either was framed as a pregnancy risk, so their answers were more likely to be candid. And the grandchildren’s diagnoses arrived 30 years or more later, when no one’s memory could have been colored by the result.

The limits of the evidence

Only 86 grandchildren in the cohort had an ASD diagnosis (72 boys, 14 girls), enough to detect an association, but not enough to draw firm conclusions about paired or multiple grandparent exposures, or about whether effects travel through the mother’s line or the father’s. State records on disability services mostly capture children needing more support, so milder autism is likely undercounted. Grandfathers’ exposures were reported second-hand by their wives, and some of that data was missing, though the authors tested for distortion and found their conclusions held. The association could be due to other factors that were not measured in the study.  Importantly, the findings describe risk across a population, not the cause of any individual child’s autism.

The takeaway

For now, associations can provide precautionary guidance. “The practical implication is that men planning to have children may have reason to quit smoking and drinking,” Cohn said.

The larger promise is scientific. If researchers can identify biological markers left behind by a grandparent’s exposures, they may be able to trace how the risk is passed down and find points where it can be interrupted.

Study: Pearl, M., K. Berger, P. M. Cirillo, et al. 2026. “Grandparent Prenatal Tobacco, Alcohol, Coffee, and Tea Consumption and Autism in the Third Generation.” Autism Research e70344. https://doi.org/10.1002/aur.70344

Funding: This work was supported by the Office of the Assistant Secretary of Defense for Health Affairs under Autism Research Program Award No. W81XWH-16-1-0254, and by the Escher Fund for Autism.