Autism is primarily genetic, but environmental factors play a meaningful supporting role. The largest study on the question, a multinational analysis of over 2 million people across five countries, estimated autism’s heritability at roughly 81%. That leaves about 18% of the variation in autism risk attributable to environmental influences, nearly all of which are unique to the individual rather than shared within a family. In short, genes do the heavy lifting, but the environment can tip the balance.
How Much Is Genetic vs. Environmental
The 80/20 split comes from a study published in JAMA Psychiatry that tracked children born in Denmark, Finland, Sweden, Iceland, and Israel. Across all five countries, inherited genetic factors consistently explained the majority of autism risk, with country-specific estimates ranging from about 51% in Finland to 87% in Israel. The Nordic countries combined landed at 81%.
Shared environment, meaning factors that siblings in the same household experience together (neighborhood, family income, household diet), accounted for almost nothing: 0.3%. Maternal effects during pregnancy contributed a similarly tiny fraction, around 0.5%. The remaining roughly 18% fell under “nonshared environment,” which includes everything from random biological variation during fetal development to individual-level exposures like infections, medications, or chemical contact. This tells us that when environment does matter, it’s not about the home you grew up in. It’s about what happened to you specifically, often before you were born.
Prenatal Exposures With the Strongest Evidence
Almost all confirmed environmental risk factors act during pregnancy, when the developing brain is most vulnerable. The clearest example is valproate, an anti-seizure medication. Children exposed to it in the womb had an absolute autism risk of about 4.4%, roughly triple the risk of unexposed children even after accounting for the mother’s epilepsy. Among children born to mothers with epilepsy specifically, valproate exposure nearly tripled the risk of childhood autism compared to unexposed children of mothers with the same condition. This is one of the most well-documented environmental contributors to autism and is a major reason doctors now avoid prescribing valproate to women who may become pregnant.
Air pollution is another area of growing concern. A large retrospective study of nearly 319,000 mother-child pairs in Southern California found associations between prenatal fine particulate matter (PM2.5) exposure and autism diagnosis by age five. The risk appeared strongest during the first two trimesters, up to about 27 weeks of pregnancy. The effect sizes are smaller than with medication exposures, but given how widespread air pollution is, even a modest increase in risk can matter at a population level.
Maternal Health During Pregnancy
The mother’s own health creates the biochemical environment the fetus develops in, and certain conditions appear to shift autism risk. Research from Johns Hopkins found that children born to mothers who were both obese and diabetic were more than four times as likely to be diagnosed with autism compared to children of healthy-weight mothers without diabetes. Gestational diabetes that developed during pregnancy carried a similar elevated risk when combined with obesity.
Infections during pregnancy also matter, though not because of any specific virus or bacterium. What seems to drive the risk is the immune response itself. When a pregnant person’s immune system mounts a strong inflammatory response, signaling molecules (particularly one called interleukin-6) can cross the placenta and disrupt the tightly choreographed sequence of fetal brain development. Animal studies have shown that this maternal immune activation alters how the placenta handles serotonin, sending abnormal amounts to the fetal brain. The timing, intensity, and genetic background of both mother and child all influence whether this immune disruption actually results in autism or another neurodevelopmental outcome.
Folic Acid and Protective Factors
Not all environmental influences increase risk. Prenatal folic acid supplementation is one of the few factors consistently linked to lower autism rates. Women who took folic acid (around 600 micrograms) in the first month of pregnancy had children roughly half as likely to receive an autism diagnosis. A large meta-analysis found that folic acid or multivitamin use before pregnancy was associated with a 61% risk reduction, and use during pregnancy with a 73% reduction.
The relationship isn’t perfectly straightforward. One study found a U-shaped pattern with multivitamins: moderate supplementation (three to five times per week) was protective, while both very low and very high intake were associated with increased risk. Overall, though, the evidence leans strongly toward a benefit, and folic acid supplementation in the first month of pregnancy was also linked to less severe autism symptoms and better language development in offspring at age three.
How Environment Changes Gene Expression
The old framing of “genes vs. environment” is increasingly outdated. What researchers now see is that environmental exposures can chemically modify how genes behave without changing the DNA sequence itself. This process, called epigenetics, works like a dimmer switch: it can turn genes up or down based on outside signals.
Certain industrial chemicals, particularly plasticizers like BPA and flame retardants, have been shown to alter these chemical tags on DNA in fetal brain tissue and cord blood. Tobacco smoke exposure during pregnancy changes gene activity in ways that are still detectable in children’s blood at age 17, suggesting these modifications are not temporary. Maternal stress and poor nutrition can do the same, affecting genes involved in stress response and energy metabolism. Even calorie or protein restriction during pregnancy can reprogram fetal gene expression to favor energy storage, a survival adaptation that may come with neurodevelopmental trade-offs.
This is likely why two children with similar genetic profiles can have very different outcomes. Their genes may be the same, but the environmental instructions layered on top of those genes during critical windows of development are not.
Parental Age
Parental age at conception is sometimes classified as an environmental factor because it reflects accumulated biological changes rather than inherited DNA. Interestingly, in families already at high genetic risk for autism, one study found that fathers under 30 had nearly three times the odds of having a child with autism compared to fathers aged 30 to 34. Fathers 40 and older showed a modest, non-significant increase (1.5 times). Maternal age had little independent effect once the father’s age was accounted for. These findings run counter to the common assumption that older parents always carry higher risk, at least within high-risk families, and suggest the relationship between parental age and autism is more complex than a simple “older equals riskier” rule.
Vaccines and Autism
The question of whether autism is environmental inevitably brings up vaccines. Multiple large-scale reviews by the National Academy of Sciences’ Institute of Medicine and other bodies have examined this question. For the MMR vaccine specifically, the evidence consistently shows no association with autism. For several other infant vaccines, reviewers have described the evidence as insufficient to confirm or rule out a relationship, largely because the specific studies needed to do so have not been conducted. This is a gap in evidence, not evidence of harm. No mechanism linking routine vaccines to autism has been established, and the broader scientific picture points overwhelmingly toward prenatal factors rather than postnatal exposures as the relevant environmental window.