Autism spectrum disorder arises from a combination of genetic susceptibility and environmental influences acting during brain development, with no single cause identified for most cases. Twin studies consistently show that genetics accounts for the majority of risk, with heritability estimates ranging from roughly 64% to 95% depending on the study and how autism is measured. But heritability is not destiny, and the remaining risk comes from a web of prenatal and perinatal factors that researchers are still untangling. The picture that has emerged over the past two decades is one of staggering complexity, where hundreds of genes, dozens of environmental exposures, and the timing of both converge to shape whether and how autism develops.
Genetics Carries the Largest Share of Risk
The clearest evidence for genetics comes from twin studies. Identical twins share virtually all their DNA, while fraternal twins share about half. If autism were purely environmental, both types of twins would be affected at similar rates. Instead, the concordance gap is enormous. A meta-analysis of twin studies found that identical-twin correlations for autism were near-perfect at about 0.98, while fraternal-twin correlations were much lower, in the range of 0.53 to 0.67. From those figures, the researchers estimated heritability at 64% to 91%.1PubMed Central. Heritability of autism spectrum disorders: a meta-analysis of twin studies A UK population-based twin sample produced a similar spread, with heritability estimates between 56% and 95% across different measures of autistic traits.2JAMA Psychiatry. Heritability of Autism Spectrum Disorder in a UK Population-Based Twin Sample
These genetic influences are also remarkably stable over childhood. A study tracking autistic traits across ages found that the genetic factors shaping those traits at one age largely overlapped with those at later ages, and that the stability of autistic traits over time was primarily driven by genetics rather than by individual environmental experiences.3PubMed Central. Strong genetic influences on the stability of autistic traits in childhood
But saying “genetics” makes it sound simpler than it is. Autism is not caused by a single gene in most cases. It is what geneticists call a complex polygenic disorder: many genetic variants, each contributing a small amount of risk, pile up together. Some of these are common variants present in a large proportion of the population, while others are rare mutations that arise spontaneously (known as de novo mutations) and can have a stronger individual effect. Research has shown that these two categories of genetic risk are not separate stories. Common polygenic risk contributes additively even in people who carry a strong-acting rare mutation.4PubMed Central. Polygenic transmission disequilibrium confirms that common and rare variation act additively to create risk for autism spectrum disorders In families with multiple autistic members, both rare inherited variants and common polygenic risk play a role, consistent with an additive architecture where different types of genetic risk stack on top of each other.5PubMed Central. The contributions of rare inherited and polygenic risk to ASD in multiplex families Overall, the common-variant genetic load appears to account for the largest portion of autism liability, with rare de novo and inherited variants acting within that broader context.6PubMed. The Yin and Yang of Autism Genetics: How Rare De Novo and Common Variations Affect Liability
Why Autism Is More Common in Males
Autism is diagnosed three to four times more often in males than females. For years, some researchers wondered whether this was simply a detection bias, since diagnostic tools were originally developed based on male presentations. But genetic evidence now strongly supports what is called a female protective effect: females genuinely require a higher genetic burden to develop autism. Siblings of autistic females have higher rates of autism than siblings of autistic males, suggesting that the families of affected girls carry, on average, a heavier genetic load.7PubMed Central. The female protective effect against autism spectrum disorder The same pattern has been replicated across large population-based samples in the UK and Sweden: when girls do cross the diagnostic threshold, their siblings show more autistic traits than the siblings of affected boys, consistent with the idea that girls need more etiological “push” to manifest the condition.8PubMed Central. Examining and interpreting the female protective effect against autistic behavior
Direct genetic comparisons reinforce this. Autistic females carry, on average, more damaging de novo mutations than autistic males. One study found that affected females had about 33% more loss-of-function mutations and 26% more damaging missense mutations than affected males.9Translational Psychiatry. Genetic evidence of gender difference in autism spectrum disorder supports the female-protective effect In other words, the bar for developing autism appears to be biologically higher for females, though the exact protective mechanism remains unclear. Mothers of autistic children also carry more common polygenic risk for autism than fathers do, adding another layer to how genetic risk is transmitted differently by sex.7PubMed Central. The female protective effect against autism spectrum disorder
Prenatal Exposures and Medications
Among environmental risk factors, one of the most well-established is prenatal exposure to valproate, a medication used primarily for epilepsy and mood disorders. A large Danish cohort study found that children exposed to valproate in utero had roughly three times the risk of autism spectrum disorder compared to unexposed children, even after adjusting for maternal epilepsy itself.10PubMed Central. Prenatal valproate exposure and risk of autism spectrum disorders and childhood autism A 2025 systematic review and meta-analysis across multiple anti-seizure medications confirmed that valproate carries the highest risk, with a hazard ratio near 2.8, while some other anti-seizure drugs like carbamazepine and oxcarbazepine showed more modest increases.11PubMed. Association of prenatal exposure to antiseizure medication with risk of autism: a systematic review and meta-analysis Not all anti-seizure medications share this risk: lamotrigine and topiramate showed no significantly elevated risk of autism in a large study published in the New England Journal of Medicine.12PubMed Central. Risk of Autism after Prenatal Topiramate, Valproate, or Lamotrigine Exposure This distinction matters for people with epilepsy who are planning pregnancies, since safer alternatives exist.
Air Pollution and Pesticides
The evidence for environmental pollutants is less dramatic in effect size but covers enormous populations, which makes even modest risk increases significant at a public health level. A systematic review and meta-analysis of air pollution studies found that prenatal exposure to fine particulate matter (PM2.5) was associated with about a 6% increase in autism risk, and exposure to nitrogen dioxide with about a 2% increase.13PubMed. Maternal exposure to air pollution and risk of autism in children: A systematic review and meta-analysis These are small per-person effects, but given how many people breathe polluted air, they could translate into a meaningful number of cases. A sibling-design study, which controls for many family-level confounders by comparing siblings exposed to different pollution levels during their respective pregnancies, confirmed the PM2.5 association and pointed to non-tailpipe sources like brake and tire wear as contributors.14PubMed. Discordant sibling analysis of autism risk associated with prenatal exposure to tailpipe and non-tailpipe particulate matter pollution
Pesticide exposure follows a similar pattern of small but consistent risk elevations. A California population-based study found modestly increased odds of autism associated with prenatal exposure to several common pesticides, including glyphosate, chlorpyrifos, and permethrin, with odds ratios in the range of 1.10 to 1.16.15PubMed Central. Prenatal and infant exposure to ambient pesticides and autism spectrum disorder in children: population based case-control study These are not large individual risks, and the studies measure proximity to agricultural pesticide use rather than direct personal exposure, so interpretation requires caution. Still, the consistency across multiple chemicals and studies suggests something real is going on.
Maternal Health During Pregnancy
Several aspects of the mother’s health during pregnancy are associated with autism risk, though disentangling these from shared genetic factors is tricky. Maternal obesity and diabetes have received considerable attention. A study published in Pediatrics found that when obesity and diabetes were examined together, the combination of pre-pregnancy obesity with either pre-existing diabetes or gestational diabetes was associated with substantially increased risk of autism in offspring, with hazard ratios around 3 to 4.16PubMed Central. The Association of Maternal Obesity and Diabetes With Autism and Other Developmental Disabilities The individual effects of obesity or diabetes alone were weaker, suggesting the combination creates a more hostile prenatal environment for brain development.
Vitamin D deficiency during pregnancy has also been linked to autism risk. A large study found that mothers with deficient vitamin D levels had about 44% higher odds of having a child diagnosed with autism compared to mothers with sufficient levels.17PubMed Central. Maternal Vitamin D Levels During Pregnancy and Offspring Autism Spectrum Disorder An independent study using the Generation R cohort in the Netherlands found a similar pattern, with mid-gestation vitamin D deficiency associated with higher autism-related trait scores, and deficient mothers being nearly four times as likely to have a child screened positive for autism-related behaviors compared to sufficient mothers.18Molecular Psychiatry. Gestational vitamin D deficiency and autism-related traits: the Generation R Study Whether supplementing vitamin D in pregnant women would actually reduce autism risk has not been established in randomized trials, and these observational associations could be confounded by factors linked to both low vitamin D and autism risk, such as reduced time outdoors or darker skin pigmentation at higher latitudes.
Parental age is another factor with solid epidemiological support. A large international study of more than five million children found that fathers aged 50 and older had about 66% higher risk of having an autistic child compared to fathers in their twenties, and that mothers aged 40 to 49 had about 15% higher risk. Interestingly, very young mothers (under 20) also showed elevated risk. The study also found that large age gaps between parents independently increased autism risk.19Molecular Psychiatry. Autism risk associated with parental age and with increasing difference in age between the parents The paternal age effect is often attributed to the accumulation of de novo mutations in sperm over time, though this likely explains only part of the association. One study in families already at high genetic risk for autism did not find a significant parental age effect, suggesting that the age-related risk may matter more in the general population than in families where genetic risk is already high.20PubMed Central. The Association Between Parental Age and Autism-Related Outcomes in Children at High Familial Risk for Autism
Maternal Immune Activation
One of the more active areas of research involves the mother’s immune system during pregnancy. The maternal immune activation hypothesis proposes that inflammatory events during pregnancy, such as severe infections, can alter fetal brain development and increase the risk of autism in the offspring.21PubMed. Maternal Immune Activation and Autism in Offspring: What Is the Evidence for Causation? Animal studies across mice, rats, and nonhuman primates have consistently shown that triggering the maternal immune response during pregnancy leads to changes in brain development and behavior in offspring, and that the immune response itself, rather than any specific pathogen, appears to be the critical factor.22PubMed Central. Maternal Immune Activation and Autism Spectrum Disorder: From Rodents to Nonhuman and Human Primates
In humans, one intriguing finding involves maternal antibodies that react against fetal brain proteins. About 8% of mothers of autistic children in one study had specific autoantibodies targeting brain proteins, and the children born to these mothers showed abnormal brain enlargement.23PubMed Central. Maternal autoantibodies are associated with abnormal brain enlargement in a subgroup of children with autism spectrum disorder This suggests a specific biological pathway that may be relevant to a subset of autism cases, not a universal mechanism.
Complications Around Birth
Preterm birth and low birth weight are consistently linked to higher autism risk, though whether they are causes or markers of the same underlying problems that led to autism is debated. Preterm birth is associated with increased risk through multiple possible pathways, including brain injury from immature blood vessels, exposure to medical interventions in neonatal intensive care, and the sheer vulnerability of a brain developing outside the womb during a critical period.24PubMed. A Systematic Review of the Risk Factors for Autism Spectrum Disorder in Children Born Preterm
The risk increase is graded: the lower the birth weight, the higher the odds. One study found that very low birth weight (under 1,500 grams) was associated with roughly three times the odds of childhood autism, while moderately low birth weight (under 2,500 grams) carried about 1.6 times the odds. Very low gestational age, under 32 weeks, showed about 2.5 times the odds.25PubMed Central. Risk of Autism Spectrum Disorders in Low Birth Weight and Small for Gestational Age Infants A large Korean study using national insurance data confirmed this dose-response pattern, with birth weights of 1.5 to 1.9 kg associated with roughly two to three times the odds of autism.26Scientific Reports. Association between birth weight and neurodevelopmental disorders assessed using the Korean National Health Insurance Service claims data
How Genes and Environment Interact
One of the most important lessons from the past decade of autism research is that genes and environment do not operate in separate lanes. The same environmental exposure can have very different effects depending on a person’s genetic background. Research into gene-environment interactions has found that the influence of environmental chemicals on autism risk likely varies depending on the genetic substrate of the exposed individual, which is part of why pinpointing specific environmental causes has been so difficult.27PubMed Central. Gene×environment interactions in autism spectrum disorders Experimental work suggests that genetic mutations and environmental toxins often disrupt overlapping regulatory pathways, meaning that a child with certain genetic vulnerabilities may be more sensitive to a particular prenatal exposure than a child without those vulnerabilities.28PubMed Central. Autism spectrum disorder at the crossroad between genes and environment: contributions, convergences, and interactions in ASD developmental pathophysiology
One way genes and environment converge is through epigenetics, where chemical modifications to DNA change how genes are expressed without altering the DNA sequence itself. DNA methylation, the most studied of these modifications, has emerged as a potential bridge between genetic predispositions and environmental exposures in autism.29Environmental Epigenetics. Intercontinental insights into autism spectrum disorder: a synthesis of environmental influences and DNA methylation A study looking at DNA methylation patterns at birth found that autism-associated methylation changes in cord blood were enriched in known autism genes, and similar enrichment was found in maternal blood and placental tissue.30PubMed Central. Autism-Associated DNA Methylation at Birth From Multiple Tissues Is Enriched for Autism Genes in the Early Autism Risk Longitudinal Investigation This suggests that some environmental influences may act on autism risk by changing how autism-related genes are turned on or off during critical windows of development.
What Happens at the Level of Brain Cells
Regardless of whether the initial trigger is genetic, environmental, or both, autism-related changes converge on how brain cells communicate with each other. Many of the genes most strongly linked to autism encode proteins that function at synapses, the junctions where nerve cells pass signals. Mutations in these synaptic proteins can disrupt both the sending and receiving sides of nerve cell communication.31PubMed. Synaptic protein mutations in autism One well-studied example involves mutations in the Shank3 protein, which not only disrupts the receiving side of the synapse but also alters signaling across the gap to change how the sending side develops and functions.32PubMed Central. Autism-associated mutations in ProSAP2/Shank3 impair synaptic transmission and neurexin-neuroligin-mediated transsynaptic signaling
Research has identified at least two clusters of genes frequently disrupted in autism: those involved in synaptic function and those involved in controlling how proteins are made within neurons. These two systems are interconnected, and mutations in either can cascade through shared downstream pathways to produce autism-related changes in brain development.33PubMed Central. Reciprocal signaling between translational control pathways and synaptic proteins in autism spectrum disorders
Another convergence point involves the balance between excitatory and inhibitory signaling in the brain. Mouse models of autism, despite having very different genetic causes, share a common reduction in a specific type of inhibitory neuron called parvalbumin-positive cells. These neurons normally drive experience-dependent circuit refinement during critical periods of brain development, and their disruption may contribute to the sensory and social difficulties seen in autism.34PubMed Central. Common circuit defect of excitatory-inhibitory balance in mouse models of autism The brain’s own immune cells, called microglia, play a role here too. These cells normally prune unnecessary synaptic connections early in life, and abnormalities in microglial function have been proposed as a contributor to several hallmarks of autism at the brain level.35PubMed Central. The Impact of Microglia on Neurodevelopment and Brain Function in Autism
The Gut and Metabolic Connections
You may have encountered claims that gut bacteria play a role in autism. There is real research behind this, though the findings are still in early stages. People with autism tend to have measurably different compositions of gut bacteria compared to non-autistic individuals, and various bacterial metabolites like short-chain fatty acids are altered in the blood and urine of autistic people.36Genes & Diseases. The microbiota–gut–brain axis and autism spectrum disorder: Microbiota-mediated mechanisms, metabolic dysregulation, and neurodevelopmental implications Whether these differences are a cause, a consequence, or a bystander effect of autism remains unclear. Dietary patterns, sensory sensitivities affecting food choices, and gastrointestinal symptoms (which are common in autism) can all alter gut bacteria independently of any causal pathway to the brain.
Similarly, research has found evidence of reduced mitochondrial function in some people with autism, including decreased activity of the cellular machinery that generates energy. A review of these studies concluded that the association is supported by available evidence, though the research base is limited by small sample sizes and methodological variability.37PubMed Central. Mitochondrial Dysfunction in Autism Spectrum Disorders As with gut bacteria, it is uncertain whether mitochondrial dysfunction is a cause of autism in some individuals, a downstream consequence of altered neurodevelopment, or a shared effect of the same genetic variants that increase autism risk.
Why Autism Diagnoses Have Increased
One question that frequently comes up in discussions of autism causes is whether something in the modern environment is actually making autism more common. The frequency of autism diagnoses has risen dramatically over several decades, but researchers disagree about how much of this reflects a true increase in incidence versus changes in how autism is recognized and diagnosed.38PubMed Central. Increasing prevalence, changes in diagnostic criteria, and nutritional risk factors for autism spectrum disorders Diagnostic criteria have broadened considerably over the years, awareness among parents and professionals has grown, and children are being evaluated earlier. A study of California’s autism caseload found that diagnostic changes alone, specifically the reclassification of individuals who previously would have been diagnosed with intellectual disability, accounted for about a quarter of the increased prevalence between 1992 and 2005.39PubMed Central. Diagnostic change and the increased prevalence of autism
Other factors likely contributing to rising diagnostic counts include the recognition that autism is a lifelong condition (leading to more adult diagnoses), earlier age of diagnosis, greater availability of services tied to a diagnosis, and heightened public awareness.40Research in Autism Spectrum Disorders. The increasing prevalence of autism spectrum disorders Whether a true increase exists on top of these ascertainment factors is genuinely unresolved. It is plausible that rising exposure to environmental risk factors like air pollution or endocrine disruptors has contributed, but separating a real signal from the noise of diagnostic expansion is one of the hardest problems in autism epidemiology.
An Evolutionary Wrinkle
Given how many genes are involved in autism and how common they are collectively, researchers have asked a natural question: why hasn’t natural selection weeded these variants out? One answer appears to be that many common genetic variants associated with autism risk show signs of having been positively selected for during human evolution. A study examining autism-associated variants found that they were significantly more likely than chance to show signatures of recent positive selection, with about a 19% increased probability of falling in the top tier of selection scores.41PLoS Genetics. Widespread signatures of positive selection in common risk alleles associated to autism spectrum disorder This implies that the same genetic variants that contribute to autism risk may have provided advantages in other contexts, perhaps in cognitive abilities, social behavior, or other traits. It also helps explain why these variants remain so common in the population despite their role in a condition that can significantly affect daily functioning.