Autism is strongly genetic. Twin studies consistently show that 64% to 91% of the variation in autism risk comes from inherited factors, with the best estimates landing around 74% to 93% depending on how broadly autism is defined. That makes autism one of the most heritable neurodevelopmental conditions, though genetics alone doesn’t tell the whole story.
What Twin Studies Reveal
The clearest evidence for a genetic basis comes from comparing identical twins (who share 100% of their DNA) with fraternal twins (who share about 50%). When one identical twin is autistic, the other twin is also on the spectrum in 60% to 90% of cases. For fraternal twins, that number drops to roughly 20%. If autism were purely environmental, identical and fraternal twins would show similar rates, since they typically share the same prenatal environment and household. The large gap between identical and fraternal concordance points squarely at genes.
A comprehensive meta-analysis of twin studies published in the Journal of Child Psychology and Psychiatry calculated overall heritability at 74%, with shared environmental factors accounting for about 25%. However, the researchers concluded that the environmental contribution was likely inflated by statistical quirks in how twin pairs were sampled. When they corrected for this, heritability rose to 93% and the environmental component became negligible. The takeaway: genetic factors are the dominant driver, though some environmental influence can’t be ruled out entirely.
There Is No Single “Autism Gene”
Autism doesn’t follow a simple inheritance pattern like eye color. Instead, two different types of genetic variation contribute to risk, and they work very differently.
The first type involves common genetic variants, tiny DNA differences that are widespread in the general population. Individually, each one nudges risk by a trivial amount. But collectively, hundreds or thousands of these small-effect variants can add up. Research estimates that common variants account for roughly half of autism’s heritability. So far, genome-wide studies have identified only five specific regions of the genome that reach statistical significance on their own, which underscores just how spread out the genetic architecture is.
The second type involves rare mutations, often spontaneous changes (called de novo mutations) that appear for the first time in a child and weren’t present in either parent. These tend to have much larger individual effects. Some disrupt genes that are critical for brain development, particularly genes that control how other genes are switched on and off during fetal growth. Researchers analyzing over 42,000 autism cases have identified a growing list of moderate-risk genes driven by these rare inherited and spontaneous variants.
About 10% of autistic individuals have a recognized single-gene condition that contributes to their diagnosis. Fragile X syndrome is the most well-known, followed by tuberous sclerosis, Rett syndrome, and PTEN hamartoma tumor syndrome. These cases are the exception, not the rule. For most autistic people, risk comes from a complex combination of many genetic variants rather than one identifiable mutation.
Recurrence Risk in Families
If you have one autistic child and are wondering about the likelihood for a younger sibling, the numbers are fairly consistent. Data from the Baby Siblings Research Consortium, which tracks younger siblings of autistic children from infancy, found that 20.2% of these siblings went on to receive an autism diagnosis. That figure hasn’t changed meaningfully since earlier estimates of 18.7% were published in 2011. For comparison, autism prevalence in the general population is roughly 2% to 3%, so siblings face about a tenfold increase in risk.
Fraternal twins of autistic children show a similar recurrence rate of about 20%, consistent with the sibling data. The sharp jump to 60% to 90% concordance in identical twins confirms that the additional shared DNA is what drives the difference, not simply growing up in the same household.
Why Autism Is More Common in Males
Autism is diagnosed in boys roughly three to four times more often than in girls. The leading explanation is sometimes called the “female protective effect.” The idea is that females, on average, require a greater number or magnitude of genetic risk factors before autism manifests. In other words, the threshold for crossing into a diagnosis is higher for girls.
Supporting this theory, studies consistently find that autistic females carry proportionally more large, highly disruptive de novo mutations than autistic males do. Among individuals whose autism is linked to a single rare, high-impact mutation, the sex ratio is close to 1:1, meaning the gap between boys and girls nearly vanishes when the genetic hit is severe enough. This suggests that many girls who carry a moderate genetic load for autism may not meet diagnostic criteria, while boys with the same genetic profile do. Whether this reflects true biological protection, differences in how autism presents in girls, or both remains an active question.
Paternal Age and New Mutations
Advanced paternal age is one of the more established non-inherited risk factors. Sperm cells accumulate new DNA mutations over a man’s lifetime at a rate of about 3.1% per year. A 45-year-old father passes on substantially more spontaneous mutations to his child than a 25-year-old father does.
In epidemiological studies, children of older fathers show roughly 68% higher odds of autism compared to children of younger fathers. Interestingly, the actual increase in risk that can be directly attributed to new mutations alone is much smaller, around 9% to 10% over that same age span. The gap suggests that paternal age affects autism risk through mechanisms beyond just accumulating random mutations, possibly through changes in how genes are regulated in sperm over time.
How Environment Interacts With Genes
Saying autism is “mostly genetic” doesn’t mean environment plays no role. The current understanding is that prenatal environmental exposures can interact with genetic susceptibility, particularly through epigenetic changes. Epigenetics refers to modifications that affect whether specific genes are active or silent without altering the DNA sequence itself.
Several genes involved in brain development, immune function, and cellular energy production show altered activation patterns in autistic individuals. Some of these changes appear to be influenced by the prenatal environment. For example, certain genes that help regulate brain cell signaling during fetal development show different activity levels in autistic versus non-autistic brain tissue. The picture that emerges is one where genetic predisposition sets the stage, and prenatal conditions can either amplify or buffer that predisposition.
This interaction also helps explain why identical twins don’t show 100% concordance. Even with the same DNA, differences in placental blood flow, exposure to maternal stress hormones, or random variation in gene activation during development can lead one twin toward a diagnosis while the other doesn’t meet criteria.
Can Genetic Testing Predict Autism?
Not reliably, at least not yet in a clinically useful way. Polygenic scores, which attempt to combine the effects of many common genetic variants into a single risk number, have shown some utility in research settings for studying groups. But systematic reviews have been clear: these scores are not accurate enough to diagnose or predict autism in any individual person. The effects of each common variant are simply too small, and the interactions between variants are too complex, for a single number to be meaningful in a clinical setting.
Genetic testing does play a role after a diagnosis, particularly for the roughly 10% of cases linked to identifiable single-gene conditions. Identifying a specific genetic syndrome can guide medical monitoring and connect families with targeted support. But for the majority of autistic individuals whose genetics involve a complex mix of common and rare variants, there is no test that can confirm or rule out autism on its own. Diagnosis remains behavioral, based on developmental history and observed patterns of communication, social interaction, and behavior.