Autism spectrum disorder does not have three neat, separate causes, but researchers consistently group the contributing factors into three broad categories: genetics, prenatal and environmental influences, and the interaction between the two. Genetics carries the heaviest weight, with heritability estimates landing around 80% or higher in large studies, but genes alone rarely tell the whole story. Prenatal exposures and complications contribute meaningfully, and the way those exposures land differently depending on a person’s genetic makeup adds a third layer that researchers are still working to untangle.
Why the “Three Causes” Framing Is Useful but Imperfect
When people search for the three main causes of autism, they are looking for a tidy breakdown. The science delivers something messier. Autism is what researchers call multifactorial: it arises from the combination of genetic and environmental factors, not from any single gene or single exposure acting alone.1PubMed Central. Autism risk factors: genes, environment, and gene-environment interactions Hundreds of genes contribute, and the relevant environmental factors range from parental age to prenatal infections to birth complications. Still, organizing these into genetics, environment, and their interaction gives a workable map of what the research actually shows. The three categories are not independent buckets so much as overlapping circles, and the edges blur more than any list of “causes” can convey.
Genetics Is the Largest Contributor
Twin and family studies have consistently identified genetics as the dominant factor. A meta-analysis of twin studies placed heritability between 64% and 91%.2PubMed Central. Heritability of autism spectrum disorders: a meta-analysis of twin studies A large population-based study across five countries estimated heritability at about 83%, with the remaining roughly 17% attributed to non-shared environmental influences.3JAMA. The Heritability of Autism Spectrum Disorder In practical terms, this means genetics explains the majority of why one person develops autism and another does not, though the genetic architecture is strikingly complex.
The genetic contribution comes in two broad flavors. The first involves rare mutations, often arising spontaneously in the child rather than being inherited from either parent. These de novo mutations are estimated to contribute to roughly 30–39% of all autism cases, and in families with no prior history of autism, that figure climbs to 52–67%.4Communications Biology. Rates of contributory de novo mutation in high and low-risk autism families A large analysis of over 42,000 autism cases found that the population-level risk attributable to damaging de novo variants is around 10%, with about two-thirds of that accounted for by already-known risk genes.5Nature Genetics. Integrating de novo and inherited variants in 42,607 autism cases identifies mutations in new moderate-risk genes
The second flavor is polygenic risk, meaning the combined effect of many common genetic variants, each contributing a tiny nudge toward or away from autism. No single common variant has a large effect on its own, but collectively they add up. A meta-analysis found that a polygenic score for autism is associated with an autism diagnosis, though the overall effect size is modest.6PubMed Central. Systematic Review and Meta-Analysis: Phenotypic Correlates of the Autism Polygenic Score In multiplex families where more than one child is autistic, research has found that autistic children tend to inherit higher polygenic risk from their non-autistic parents, and this inherited common-variant risk can combine with rare inherited variants to push a child over a threshold that the parent never crossed.7PubMed Central. The contributions of rare inherited and polygenic risk to ASD in multiplex families
Syndromic Forms of Autism
Some genetic conditions carry an especially high rate of autism. Fragile X syndrome and tuberous sclerosis complex are the most frequently cited examples. These are sometimes called syndromic autism because the autism co-occurs with a clinically recognizable pattern of physical features and other medical issues, and each has a known single-gene cause.8PubMed Central. Syndromic autism spectrum disorders: moving from a clinically defined to a molecularly defined approach People with these syndromes who meet criteria for autism tend to show a somewhat different profile compared to those with non-syndromic autism. A cross-syndrome comparison found that individuals with fragile X had the highest overall autism severity scores and those with neurofibromatosis type 1 had the lowest, and the syndrome groups generally showed fewer social difficulties but more restricted and repetitive behaviors compared to people with non-syndromic autism.9PubMed Central. Autism Symptoms in Children and Young Adults With Fragile X Syndrome, Angelman Syndrome, Tuberous Sclerosis Complex, and Neurofibromatosis Type 1: A Cross-Syndrome Comparison These syndromic forms account for a small fraction of all autism cases, but they have been enormously valuable for understanding the biology because studying a known genetic mutation is far easier than tracing the effects of hundreds of common variants at once.
Prenatal and Environmental Risk Factors
The second major category encompasses everything that happens outside the genome itself, particularly during pregnancy and around the time of birth. The word “environmental” here does not mean pollution alone; in research terms, it covers any non-genetic influence, from parental age to infections to medication use to birth complications.
Parental age is one of the most replicated findings. A large international study found that mothers aged 40–49 had about a 15% higher risk of having an autistic child compared to mothers aged 20–29, and fathers aged 50 or older had roughly 66% higher risk compared to fathers in their twenties, after adjusting for the other parent’s age.10Molecular Psychiatry. Autism risk associated with parental age and with increasing difference in age between the parents An earlier study found broadly consistent results, with maternal age 35 and older and paternal age 40 and older both independently linked to increased odds of autism.11American Journal of Epidemiology. Advanced Parental Age and the Risk of Autism Spectrum Disorder The mechanism likely involves the accumulation of spontaneous mutations in sperm and eggs over time, which connects the “environmental” risk of older parental age back to the genetic category in an interesting way.
Maternal infections during pregnancy have received significant attention. Animal studies and human epidemiological data both point to maternal immune activation as a risk factor. When the mother’s immune system mounts a strong response to an infection, the resulting inflammatory signals can affect fetal brain development, even without the infectious agent itself crossing the placenta.12PubMed Central. Maternal Infection during Pregnancy and Autism Spectrum Disorders A study using genetic and clinical data found that maternal pregnancy-related infections were associated with about a 34% increased odds of autism in the offspring.13Translational Psychiatry. Maternal pregnancy-related infections and autism spectrum disorder—the genetic perspective
Certain medications taken during pregnancy also carry risk. The anti-epilepsy drug valproate is the clearest example. A Danish national cohort study found that children exposed to valproate prenatally had an absolute risk of about 4.4% for autism spectrum disorder, with nearly a threefold increase compared to unexposed children, even after adjusting for the mother’s epilepsy.14PubMed Central. Prenatal valproate exposure and risk of autism spectrum disorders and childhood autism Valproate is now avoided during pregnancy whenever alternatives exist, but for some women with epilepsy it remains the only effective option, creating a genuine clinical dilemma.
Birth complications add another layer. A comprehensive meta-analysis identified a long list of perinatal factors associated with autism risk, including abnormal fetal presentation, umbilical cord complications, fetal distress, low birth weight, low Apgar scores, and multiple birth.15Pediatrics. Perinatal and Neonatal Risk Factors for Autism: A Comprehensive Meta-analysis Preterm and early-term birth have also been linked to increased autism risk in both boys and girls, with analyses suggesting the association is largely independent of shared familial factors.16PubMed Central. Preterm or Early Term Birth and Risk of Autism None of these complications are strong enough individually to “cause” autism in the way a single gene mutation can, but they add cumulative stress during a period of rapid brain development.
Environmental pollutants round out the category. A systematic review concluded that exposure to metals, pesticides, and air pollutants during early life is a risk factor for neurodevelopmental disorders including autism.17PubMed. The contribution of environmental pollutants to the risk of autism and other neurodevelopmental disorders: A systematic review of case-control studies A population-based California study found that prenatal proximity to several common pesticides, including glyphosate and chlorpyrifos, was associated with modestly elevated odds of autism.18PubMed Central. Prenatal and infant exposure to ambient pesticides and autism spectrum disorder in children: population based case-control study The effect sizes were small, around 10–16% increased odds per pesticide, and these are observational findings that cannot prove causation on their own. But they fit a broader pattern suggesting that chemical exposures during critical windows of fetal brain development can nudge risk.
Gene-Environment Interactions as the Third Factor
The third category is harder to pin down because it is not a separate set of causes so much as a description of how the first two interact. The same environmental exposure can have very different effects depending on a person’s genetic makeup. Research on how environmental chemicals interact with autism genetic susceptibilities is a growing area, driven by the recognition that the influence of chemical exposures likely varies depending on the genetic background of the individual.19PubMed Central. Gene×environment interactions in autism spectrum disorders
One concrete mechanism for this interaction is epigenetics, where environmental conditions change how genes are read without altering the DNA sequence itself. DNA methylation, a chemical tag that can switch genes on or off, has drawn particular interest. Evidence for abnormal DNA methylation in autism has been found at multiple levels, and critically, the period of greatest vulnerability for these changes aligns with embryonic development and early postnatal life, exactly when synapse formation is at its peak.20PubMed Central. DNA Methylation and Susceptibility to Autism Spectrum Disorder This means an environmental exposure during a specific developmental window could leave a lasting mark on gene activity in ways that neither the genes alone nor the exposure alone would predict.
What Happens in the Brain
Regardless of whether the initial trigger is genetic, environmental, or both, many lines of evidence converge on synaptic dysfunction as a shared downstream pathway. Mutations in genes associated with autism tend to cluster in pathways that control how synapses form, are pruned, and transmit signals. These genes encode molecules involved in cell adhesion, scaffolding, and protein production at the synapse, suggesting that autism involves disrupted communication between neurons at a fundamental level.21PubMed Central. Synaptopathology Involved in Autism Spectrum Disorder
Brain growth patterns also tell a story. Research has documented abnormal brain overgrowth during the first two years of life in children later diagnosed with autism, particularly in areas involved in social, emotional, and language function. This rapid overgrowth is followed by a period of abnormally slow or arrested growth.22PubMed. Brain development in autism: early overgrowth followed by premature arrest of growth The timing matters: this overgrowth happens exactly when the brain is building its most complex circuits, and disruption during this window can lead to the aberrant connectivity that underlies autistic traits. Proteomic studies of the cerebellum in autism support the idea that these structural and functional abnormalities begin during fetal development and evolve across childhood into adulthood.23PubMed Central. Impaired Aggrephagy, Interrupted Vesicular Trafficking, and Cellular Stress, Lead to Protein Aggregation, and Synaptic Dysfunction in Cerebellum of Children and Adults with Idiopathic Autism
The Vaccine Myth and Why It Persists
No discussion of autism causes would be complete without addressing the claim that vaccines, or the thimerosal preservative once used in some childhood vaccines, cause autism. The evidence here is unambiguous. A critical review of epidemiological studies found that the preponderance of evidence does not support any association between thimerosal-containing vaccines and autism, and that the studies claiming to show a link were of poor quality and could not be meaningfully interpreted.24Pediatrics. Thimerosal-Containing Vaccines and Autistic Spectrum Disorder: A Critical Review of Published Original Data In 2004, the Institute of Medicine formally rejected the hypothesis of a causal relationship.25PubMed Central. Thimerosal-containing vaccines and autism: a review of recent epidemiologic studies
The myth endures partly because of timing. Autism symptoms often become noticeable around the same age that children receive routine vaccinations, creating the illusion of a connection. The rise in autism diagnoses over the past few decades has fed the misperception, but that rise is largely explained by factors that have nothing to do with biology. Broadened diagnostic criteria, the inclusion of milder presentations, and an earlier average age of diagnosis each contributed to the increase, though one analysis found that these factors together may not fully account for the entire magnitude of the rise.26PubMed Central. The Rise in Autism and the Role of Age at Diagnosis Earlier epidemiological work concluded that the majority, if not all, of the increase in reported prevalence was due to these shifting diagnostic and awareness patterns.27PubMed. The epidemiology of autistic spectrum disorders: is the prevalence rising? Whether a small residual increase reflects genuine changes in environmental exposures or simply further diagnostic expansion remains an open question, but vaccines have been ruled out as a contributor by every credible line of investigation.
Why Boys Are Diagnosed More Often Than Girls
Autism is diagnosed in boys roughly three to four times more often than in girls, and one leading explanation is what researchers call the female protective effect. The idea is that females require a greater genetic load to manifest autism. Supporting this, siblings of autistic girls have higher rates of autism than siblings of autistic boys, suggesting that girls who do develop autism carry more familial risk and pass more of it to relatives.28PubMed Central. Examining and interpreting the female protective effect against autistic behavior A study using Danish and U.S. datasets found that mothers of autistic children carried more common-variant polygenic risk for autism than fathers did, and that unaffected male siblings under-inherited that polygenic risk, both consistent with a protective factor in females.29PubMed Central. The female protective effect against autism spectrum disorder
That said, the picture is not fully settled. A detailed examination of the predictions that should follow from the standard liability threshold model found that several key predictions were not actually supported by current data, indicating that the female protective effect may not work in the straightforward way the leading model assumes.30PubMed Central. Can the “female protective effect” liability threshold model explain sex differences in autism spectrum disorder? The sex difference is real and robust, but the explanation is probably more complicated than a simple threshold shift. Diagnostic bias also plays a role: girls with autism are more likely to be missed or diagnosed late, particularly when their presentation involves fewer stereotypical repetitive behaviors and more internalizing difficulties like anxiety. How much of the sex ratio reflects genuine biological protection versus underdiagnosis in girls remains an active debate.
Polygenic Risk and the General Population
An underappreciated finding is that the common genetic variants associated with autism do not only exist in autistic people. They are scattered across the entire population. A study using brain imaging found that higher polygenic scores for autism were associated with reduced neurite density in both children and adults from the general population, a result consistent across the cortex and white matter.31Molecular Psychiatry. Polygenic scores for autism are associated with reduced neurite density in adults and children from the general population This tells us that the genetic variants linked to autism influence brain structure on a continuum, not just in people who cross a diagnostic threshold. Autism-associated polygenic risk has also been linked, in systematic review, to social behavior, depression, and motor skills in the broader population, though most of these effect sizes are very small.6PubMed Central. Systematic Review and Meta-Analysis: Phenotypic Correlates of the Autism Polygenic Score
This continuum perspective has practical implications. It means autism is not a binary switch flipped by a specific cause. Instead, every person sits somewhere on a distribution of genetic risk, and prenatal or postnatal events shift their position on that distribution. For most people, the total load of genetic and environmental risk stays well below any clinical threshold. For a smaller number, multiple factors align to produce the behavioral and cognitive profile we recognize as autism. The boundaries are drawn by diagnostic convention, not by a sharp biological dividing line.
What Evolutionary Biology Adds to the Picture
If so many genes contribute to autism risk, and those genes are common in the population, a natural question is why natural selection has not weeded them out. One hypothesis proposes that some autism-associated gene variants were actually preserved because they conferred adaptive advantages in certain ancestral environments. The solitary forager hypothesis, for instance, suggests that traits like focused attention, systematic thinking, and comfort with solitude could have been advantageous for individuals who foraged alone rather than in groups.32PubMed Central. Conceptualizing the autism spectrum in terms of natural selection and behavioral ecology: the solitary forager hypothesis This is speculative and difficult to test directly, but it offers a framework for understanding why autism-related genetic variation is so widespread. The traits that become disabling in modern social environments may have been neutral or beneficial in different ecological contexts. Whether or not specific evolutionary scenarios prove correct, the persistence of these variants across human populations strongly suggests they are not purely detrimental from a genetic fitness standpoint.