Autism runs through twin pairs at strikingly high rates compared to ordinary siblings, and the difference between identical and fraternal twins is one of the strongest pieces of evidence that autism has a large genetic component. When one identical twin has autism, the other twin also meets diagnostic criteria roughly 80 to 96 percent of the time, depending on the study and how strictly diagnosis is applied. For fraternal twins, that figure drops to somewhere between 20 and 40 percent. These numbers have made twin research central to our understanding of autism’s origins, but the picture gets more interesting once you look beneath the headline concordance rates.
How Often Both Twins Are Diagnosed
The concordance rate, meaning how often both twins in a pair share a diagnosis, is the core measurement in autism twin research. A meta-analysis pooling data from multiple twin studies found that the correlation in autism traits for identical twins was almost perfect, at .98.1PubMed Central. Heritability of autism spectrum disorders: a meta-analysis of twin studies A separate quantitative analysis reported probandwise concordance for autism diagnosis at 96 percent in identical twin pairs.2PubMed Central. On the Nature of Monozygotic Twin Concordance and Discordance for Autistic Trait Severity: A Quantitative Analysis These are remarkably high numbers, and they tell a clear story: if one identical twin is autistic, the other almost certainly will be too.
Fraternal twins share about half their genes on average, which is the same as any pair of siblings. Their concordance rates reflect that reduced genetic overlap. A study of 277 twin pairs found that male-male fraternal pairs were about 40 percent concordant for autism, while fraternal pairs that included at least one female were only about 20 percent concordant.3JAMA Pediatrics. Characteristics and Concordance of Autism Spectrum Disorders Among 277 Twin Pairs That gap between identical and fraternal twins is one of the clearest signals in behavioral genetics that genes are doing heavy lifting.
What Heritability Actually Means Here
Twin studies are the primary tool researchers use to estimate heritability, which is the proportion of variation in a trait that can be attributed to genetic differences in a population. For autism, most estimates cluster around 80 to 90 percent. A large reanalysis of familial risk data estimated heritability at about 83 percent, with shared environmental factors contributing very little.4JAMA. The Heritability of Autism Spectrum Disorder That aligns with earlier twin studies that put the figure around 90 percent.
One well-known California twin study threw a wrench into this consensus by estimating genetic heritability at only about 37 to 38 percent and attributing a large share, around 55 to 58 percent, to shared environmental factors.5PubMed Central. Genetic heritability and shared environmental factors among twin pairs with autism That study made headlines because it suggested the environment might matter far more than previously thought. But the confidence intervals were wide, and later reanalyses with larger datasets pushed the heritability estimate back up.4JAMA. The Heritability of Autism Spectrum Disorder The current scientific consensus leans toward genetics explaining most of the risk, though no serious researcher thinks genes are the whole story.
When researchers look at autistic traits in the general population rather than just clinical diagnoses, the genetic contribution is moderate but still substantial. One twin study of autistic traits found that additive genetic influences accounted for about 57 percent of the variation, with the rest explained by environmental influences unique to each individual.6JAMA Pediatrics. Heritability of Autistic Traits in the General Population This matters because it suggests the genetics of autistic traits, as measured on a spectrum in everyone, may work somewhat differently from the genetics of clinical-level autism.
Why Identical Twins Can Still Differ
If identical twins share virtually all their DNA, you might expect them to always match on an autism diagnosis. And they usually do. But the roughly 4 to 15 percent of identical pairs who are discordant, meaning one has a clinical diagnosis and the other does not, have attracted intense scientific interest. These pairs are a natural experiment for identifying what factors beyond DNA sequence affect whether someone develops autism.
The most important concept here is the non-shared environment: experiences and exposures that differ between two siblings even when they grow up in the same household. For twins, non-shared factors can start before birth. Differences in placental blood supply, positioning in the womb, and random variation in early brain development all count. A systematic review of discordant identical twin pairs found that the non-shared environment in autism may be linked to a range of biological mechanisms that arise after the fertilized egg splits, including mutations that occur in one twin but not the other.7PubMed Central. Behavioral and biological divergence in monozygotic twin pairs discordant for autism phenotypes: A systematic review
One research team used whole-genome sequencing to look for genetic differences between identical twins where only one had autism. They identified specific mutations present in the affected twin but not the unaffected one, pointing to new candidate regions in the genome that may contribute to autism risk.8PubMed Central. Identifying Genomic Variations in Monozygotic Twins Discordant for Autism Spectrum Disorder Using Whole-Genome Sequencing These post-fertilization mutations are sometimes called somatic mutations, and they remind us that “identical” twins are not perfectly identical at the molecular level.
There is also an important nuance in how we define discordance. Even when both identical twins meet diagnostic criteria, they can differ substantially in the severity of their autistic traits. The same quantitative analysis that reported 96 percent concordance for the diagnosis itself found that, among twins who both cleared the diagnostic threshold, the correlation in trait severity was surprisingly low, on the order of 0.1.2PubMed Central. On the Nature of Monozygotic Twin Concordance and Discordance for Autistic Trait Severity: A Quantitative Analysis In other words, genetics might determine whether you cross the line into a clinical diagnosis, but how far past that line you go seems to be shaped more by non-genetic factors. That finding has practical implications: it suggests that identifying and targeting those non-shared environmental influences could help with early intervention strategies.
Prenatal Factors That Do and Don’t Matter
Parents of twins sometimes wonder whether specific pregnancy complications might explain why one twin developed autism and the other did not. A twin study that examined a range of prenatal and perinatal risk factors, including parental age, prenatal medication use, uterine bleeding, and premature birth, found that none of these shared factors increased concordance for autism in twin pairs.9PubMed Central. Prenatal and Perinatal Risk Factors in a Twin Study of Autism Spectrum Disorders That makes sense when you think about it: these are exposures that affect both twins similarly, so they would not explain why one twin is autistic and the other is not. They might contribute to baseline risk, but they do not explain within-pair differences.
A separate concern sometimes raised is whether in vitro fertilization (IVF) and other assisted reproductive technologies, which are more common in twin pregnancies, raise autism risk. A large study found that the small risk associated with IVF procedures was not explained by parental age or parental psychiatric history.10JAMA. Autism and Mental Retardation Among Offspring Born After In Vitro Fertilization The relationship between IVF and autism remains an area of active research, but the effect sizes are small, and it is difficult to separate the procedure itself from the underlying fertility issues that led to IVF in the first place.
Sex Differences in Twin Concordance
Autism is diagnosed more often in boys than girls, and this sex imbalance shows up clearly in twin concordance data. In the study of 277 twin pairs mentioned earlier, female identical twins were 100 percent concordant, meaning every pair where one girl was autistic, the other was too. Male identical twins were 86 percent concordant, which is still very high but leaves some room for discordance.3JAMA Pediatrics. Characteristics and Concordance of Autism Spectrum Disorders Among 277 Twin Pairs
The fraternal twin data were even more striking. Male-male fraternal pairs had about 40 percent concordance, while fraternal pairs with at least one female member had only about 20 percent concordance. Having at least one girl in a fraternal pair cut the risk of concordance roughly in half.3JAMA Pediatrics. Characteristics and Concordance of Autism Spectrum Disorders Among 277 Twin Pairs This pattern is consistent with what researchers call the “female protective effect,” the idea that girls need a higher genetic load to develop autism than boys do. When a girl does develop autism, it suggests a particularly strong genetic influence, which is why her identical twin is almost always affected too.
The Broader Autism Phenotype in Co-Twins
When researchers find an identical twin pair where one has autism and the other does not, they often discover that the unaffected twin is not entirely typical either. In studies of identical twins discordant for autism, more than 75 percent of the co-twins without a full autism diagnosis still displayed what is called the broader autism phenotype: subclinical traits like social aloofness, rigid thinking, or difficulty with pragmatic language.11American Journal of Psychiatry. Neuroanatomic variation in monozygotic twin pairs discordant for the narrow phenotype for autism This means true discordance, where one twin is fully autistic and the other shows absolutely no related traits, is quite rare.
This broader phenotype finding reinforces the genetic picture. The shared DNA pushes both twins toward autistic traits, but additional factors, whether somatic mutations, subtle differences in prenatal development, or random variation in brain wiring, determine whether someone crosses the clinical threshold. It also has implications for family members beyond twins. Parents and siblings of autistic individuals are more likely to show broader phenotype traits themselves, which makes sense given the high heritability.
How the Diagnostic Manual Changes the Numbers
Concordance rates are not fixed biological constants. They depend heavily on how autism is defined and measured, and those definitions have shifted over time. A study that applied both the older DSM-IV and the newer DSM-5 criteria to the same group of twin pairs found substantially different concordance rates depending on which manual was used. Under strict diagnostic methods, concordance was about 57 percent using the older criteria but dropped to about 36 percent using DSM-5 criteria. With more relaxed assessment methods, the rates were closer: about 86 percent for the older criteria and 79 percent for the newer ones.12Research in Autism Spectrum Disorders. ASD concordance of twins across DSM-IV-TR and DSM-5 diagnostic criteria
These swings are a reminder that “how common is autism in twins” does not have a single fixed answer. The strictness of the diagnostic tool, the version of the diagnostic manual, and whether you rely on parent interviews, clinician observation, or both can all shift the resulting concordance rate by a wide margin. When comparing numbers across different studies, the assessment method often matters as much as the biology.
Brain Differences in Discordant Twin Pairs
Neuroimaging studies of twin pairs where one has autism and the other does not have revealed that the brain differences are not as clean-cut as a simple “autism brain versus normal brain” distinction. In one study, both twins in discordant pairs, the one with autism and the one without, had lower frontal, temporal, and occipital white matter volumes compared to unrelated comparison subjects.13PubMed. Neuroanatomic variation in monozygotic twin pairs discordant for the narrow phenotype for autism In other words, both twins showed some brain differences linked to familial risk, even though only one had a clinical diagnosis. The twins differed from each other mainly in cerebellar volumes, with concordant pairs matching on cerebellar gray and white matter but discordant pairs diverging there.
A detailed case study of one discordant identical pair found that the affected twin had markedly smaller caudate, amygdala, and hippocampal volumes compared to his brother. Yet both twins showed reduced volumes in the superior temporal gyrus and frontal lobe relative to unrelated controls.14PubMed. Neuroanatomical and neurocognitive differences in a pair of monozygous twins discordant for strictly defined autism The researchers interpreted this as evidence for two overlapping brain networks: one subcortical network that distinguished the affected twin from his brother and may underlie core autism features, and a second cortical network that distinguished both twins from the general population and may relate to the broader phenotype.
Research on cortical folding patterns has added another layer. In identical twin pairs varying in autism discordance, folding patterns across most brain regions were highly discordant within pairs, suggesting that the fine-grained structural architecture of the cortex is shaped significantly by non-genetic factors.15PubMed. Gyrification patterns in monozygotic twin pairs varying in discordance for autism More recent work using connectivity analysis found that twins with clinical autism had reduced connectivity between the brainstem and visual cortex compared to their co-twins, while twins with higher autistic traits (even below the diagnostic threshold) showed reduced connectivity in networks involving the hippocampus and the fusiform gyrus, an area important for face processing.16Scientific Reports. Reduced brain connectivity along the autism spectrum controlled for familial confounding by co-twin design
Together, these imaging studies paint a consistent picture: shared genetics create a common backdrop of brain differences in both twins, but the specific configuration of subcortical structures, connectivity patterns, and cortical folding varies enough between twins to push one past the diagnostic line and leave the other below it.
Does Being a Twin Itself Raise Autism Risk?
One persistent question is whether the twinning process itself, whether sharing a womb, being born prematurely (which is more common in twin pregnancies), or other twin-specific factors, increases the likelihood of autism. The evidence says no. A population-based study from Western Australia found that the rate of twin births among children diagnosed with autism was about 30 per 1,000, which was not meaningfully higher than the overall twin birth rate in the population of roughly 26 per 1,000.17PubMed Central. On the twin risk in autism The study concluded that the high proportion of twins seen in some autism research samples is explained by the high concordance rate in identical twins and by how families are recruited into studies, not by twinning itself being a risk factor.
This is worth knowing because twins do face higher rates of certain complications like low birth weight and preterm delivery, and those complications are sometimes loosely associated with developmental outcomes. But the data do not support a direct path from being a twin to being more likely to develop autism. The twin-specific complications do not appear to add meaningful risk beyond what genetics already explains.
What Families of Twins Should Know
For families who have one twin diagnosed with autism, the research carries some practical implications. If the twins are identical, the probability that the second twin also meets criteria is very high, and screening early rather than waiting for obvious signs makes sense. Even if the second twin does not receive a full diagnosis, the broader phenotype data suggest that subclinical traits are common and may benefit from support. If the twins are fraternal, the risk is lower but still well above the general population rate, particularly for boy-boy pairs. Because diagnostic criteria and assessment methods affect the numbers so substantially, getting a thorough evaluation from a clinician experienced with autism, rather than relying on a single screening questionnaire, matters for both twins.
The severity question is also relevant for families. Even when both identical twins are autistic, their day-to-day profiles can look quite different. One may need significant support while the other manages more independently. The research on trait severity within diagnosed pairs suggests that the non-shared environment plays a large role in shaping those differences, which means that individualized approaches to education, therapy, and support are warranted even when both twins carry the same diagnosis.