Heritability and families
How twin and family studies estimate the heritability of autism, what that figure does and does not mean, and what autistic people and their families can take from it.
How scientists work out how much of autism comes from genes, what that number really means, and what families can expect.
What "heritability" measures Established
Heritability is the share of the differences between people in a trait, across one population at one time, that is linked to genetic differences.1 It is not a fixed property of autism, and it is not a person's chance of being autistic or of passing autism on.2 When a study reports that autism is highly heritable, it means that in that group, under the environments those people lived in, genetic differences accounted for much of the variation in who is autistic.1,3 Change the population or the environment and the number can change.1
Heritability is a way to talk about a whole group of people. It is the share of the differences between them that comes from their genes.1 It is not one person's chance of being autistic. It is also not a person's chance of passing autism on.2 The number can change if the group or the world around them changes.
The definition comes from a standard methods review, and the way the figure is reported in autism research is consistent across the large studies below.
The share of the variation between people in a trait that is linked to genetic differences, measured across a population.
Scientists agree on what this word means. Many big studies use it.
The share of the differences between people in a group that comes from their genes.
Twin and family studies Established
Twin studies compare identical twins, who share nearly all their DNA, with non-identical twins, who share about half. Pooling the 7 twin studies that met a systematic recruitment standard, out of 13 identified, identical twins' autism measures correlated almost perfectly (0.98) and non-identical twins' much less (0.53 to 0.67, depending on the assumed prevalence), giving a heritability of about 64% to 91%.3 A single UK twin cohort gave 56% to 95% across different measures, with the higher figures resting on small diagnostic subsamples.4
Whole-population register studies give similar figures. A five-country analysis of about 2 million children put the median heritability at 80.8%, with country estimates from about 51% to 87% across different models.5 An earlier analysis of Swedish national registers, covering children born from 1982, gave a heritability of about 50% in a model that kept only additive genetic and nonshared environmental effects.2 A California twin study of 192 pairs estimated heritability at only 37% to 38%, with a large share attributed to shared environment, but its confidence intervals were very wide (as low as about 8%); larger register analyses have not found a shared-environment effect.6,3,5 The spread shows how much the estimate depends on the population, the definition of autism, and the model.
Twin studies compare identical twins with non-identical twins. Identical twins share almost all their DNA. Non-identical twins share about half. When one identical twin is autistic, the other usually is too. This happens much less in non-identical twins.3 Together, 7 of the 13 twin studies found a figure of about 64% to 91%.3 A UK study found 56% to 95%.4
Whole-country studies find similar numbers. One study of about 2 million children found a figure near 81%.5 An earlier study of Swedish records gave about 50% with a different method.2 A California study found only about 38%, but its numbers were very uncertain.6 The number depends on the method.
Large twin studies and whole-population register studies find a substantial genetic contribution to autism. The exact figure moves with the population, the definition of autism, and the model; the lowest estimates come from single twin studies with very wide confidence intervals.
Big twin studies and whole-country studies agree that genes matter a lot. The exact number changes with the group and the way autism is counted.
Common and rare variation Established
Autism is polygenic: it is influenced by many common variants, each with a small effect, rather than by one or a few genes. In one Swedish sample, most of the genetic risk lay in common variation, with a narrow-sense heritability of about 52%. Rare de novo (new) mutations matter greatly for the person who carries one, but they account for only about 2.6% of the variation in the population.7 A large genome-wide association study found five significant loci and confirmed that common variants contribute substantially.8
Rare, high-impact mutations are found in a minority of autistic people. In a large exome study, de novo loss-of-function mutations in a set of 107 genes enriched for autism risk occurred in over 5% of the autistic people who were sequenced.9 A larger joint analysis of 63,237 people, pooling rare protein-truncating and missense variants and copy-number changes, associated 72 genes with autism at a stringent threshold.10 A polygenic score built from common variants is linked to a diagnosis, but the link is weak and the score has no clinical use.11
Autism is not caused by one gene. Many common gene changes, each with a small effect, add up to most of the risk.7 Some rare changes have a big effect on one person. They are only a small part of the picture.7 One big study found five places in the DNA that are linked to autism.8
Some autistic people have a rare new gene change that switches off a gene.9 A bigger study looked at many kinds of rare change together and linked 72 genes to autism.10 A score that adds up many small changes can guess a little about autism. It is not good enough to use in a clinic.11
Two large independent analyses agree that common variation carries much of the genetic risk, and the polygenic-score finding comes from a meta-analysis.
Influenced by many genes, each with a small effect, rather than by one gene.
A genetic change that appears for the first time in a child and was not inherited from either parent.
A polygenic score combines many common variants into a single number. For autism it correlates only weakly with a diagnosis, and it is not used in clinics.11
Two big studies agree that many small gene changes add up to most of the risk.
Caused by many genes, each with a small effect, not by one gene.
A gene change that starts in a child. It does not come from either parent.
Simplex and multiplex families Emerging
Families differ in their pattern. A rough rule of thumb in the research is that families with one autistic member more often involve a new (de novo) mutation; in a study of 516 such families, autistic children carried more gene-disrupting new mutations than their non-autistic siblings.12 Families with several autistic members across generations more often carry inherited rare variants; whole-genome sequencing of 21 such families found a modest excess of rare variants in autism-associated genes.13 The split is a heuristic, not a sharp line: the sibling studies carry ascertainment, reporting and stoppage biases, such as families stopping childbearing after an autistic child.14
Clinical genetic testing looks for these causes. In an older clinic sample of 258 children from 2015, chromosomal microarray found a molecular diagnosis in 9.3%, and among the 95 children who also had exome sequencing, sequencing found a cause in 8.4%.15 In a toddler clinic, 12% of the children who were tested had a pathogenic finding.16 A negative test does not mean autism is not genetic; it often reflects inherited variants that sit below the tests' detection limits, or common variation the tests cannot read. The full account of what testing can and cannot find is in Genes and the genome.
Families are different. In some families, only one person is autistic. New gene changes are a big part of the risk here. In one study of 516 families, autistic children had more gene changes. Their non-autistic brothers and sisters had fewer.12 In other families, several people are autistic across the generations. Passed-down rare gene changes matter more here.13 This is a rough rule, not a sharp line.14
A genetic test can look for these causes. In one older clinic, a test called a microarray found a cause in about 9 in 100 children. A test that reads all the genes was done in 95 of the children. It found a cause in about 8 in 100 of them.15 In a clinic for toddlers, about 12 in 100 children who were tested had a cause found.16 If a test finds nothing, it does not mean autism is not genetic. It often means the cause is gene changes that tests cannot read yet. You can read more about testing in Genes and the genome.
The pattern rests on a small number of exome and whole-genome studies, and the clinic figures come from single samples.
A family in which more than one person is autistic.
This comes from a few studies. The clinic numbers come from single clinics.
A family where more than one person is autistic.
The broader autism phenotype Established
Many relatives of autistic people have autistic traits without meeting the criteria for a diagnosis: close attention to detail, intense interests, or a different social style. This is the broader autism phenotype.17 In parents of autistic children, the share with these traits varies widely between studies, from about 3% to 80%, depending on how the traits are measured; it is more common in fathers than mothers.18
Many relatives of autistic people have some autistic traits. They may notice small details, have strong interests, or have a different social style. They do not meet the rules for a diagnosis. This is called the broader autism phenotype.17 Among parents of autistic children, the share with these traits is very different from study to study. It is between about 3 in 100 and 80 in 100. It is more common in fathers.18
A systematic review of 41 studies and a separate review agree that sub-diagnostic traits are more common in relatives.
Autistic traits in relatives of autistic people who do not meet the criteria for a diagnosis.
Many studies agree that relatives of autistic people often have some autistic traits.
Some autistic traits in relatives who do not have a diagnosis.
The female protective effect Emerging
Autism is diagnosed more often in boys than in girls. One explanation is that girls need a larger genetic load to reach the threshold for a diagnosis.19 In two population twin samples, the siblings of girls scoring above the 90th percentile for autistic traits had more autistic traits than the siblings of boys above the same threshold.19 In genetic data, mothers of autistic children carried more autism polygenic risk than fathers, and unaffected male siblings inherited less polygenic risk than expected, though the difference was small.20 One Swedish analysis found a heritability of 87% in males (about 81% to 93%) and 76% in females (about 68% to 83%).21
The explanation usually offered, a "differing thresholds" model, has been challenged. Dougherty and colleagues (2022) argue that several of its predictions are not supported by current data.22 Girls are also diagnosed less often than boys (in the Swedish analysis, 0.8% of girls against 1.5% of boys), and these genetic designs do not separate that under-recognition from a biological protection.21
More boys are diagnosed as autistic than girls. One idea is that girls need more gene changes to be diagnosed.19 In two population twin samples, the brothers and sisters of girls with many autistic traits had more autistic traits. The siblings of boys with many autistic traits did not.19 In DNA studies, mothers of autistic children carried more autism risk than fathers. The difference was small.20 One Swedish study found a figure of 87% in males and 76% in females.21
The idea used to explain this has been questioned. Dougherty and colleagues (2022) say that some of its predictions do not fit the data.22 Girls are also diagnosed less often than boys, and these studies cannot tell that apart from a real protection.21
The finding is replicated in large samples, but the model usually offered to explain it is disputed, and the studies cannot separate a biological protection from the later and less frequent diagnosis of girls.
The idea that girls need more genetic risk than boys to be diagnosed as autistic.
Big studies agree on the pattern. But the idea used to explain it is argued about.
The idea that girls need more gene changes than boys to be autistic.
Parental and grandparental age Emerging
Older parents are slightly more likely to have an autistic child. Pooling 27 studies, the highest parental-age category had about 1.4 to 1.6 times the likelihood of the reference group, and each extra 10 years of paternal age went with about a 21% higher likelihood.23 These are associations, not evidence that age causes autism.
Grandparental age has also been studied. In Swedish registers, men who fathered a child at 50 or older had about 1.7 to 1.8 times the likelihood of a grandchild with autism, through both daughters and sons.24 A Danish study found parental age over 30 monotonically linked to autism; its grandparental findings went both ways, with higher risk for grandchildren of some young grandparents and some older grandmothers.25
Older parents are a little more likely to have an autistic child. Putting 27 studies together, the oldest parents had a higher chance. It was about 1.4 to 1.6 times the chance of the middle-age group.23 This does not prove that age causes autism.
Grandparent age has also been studied. In Swedish records, some men had a child at 50 or older. Their grandchildren were a little more likely to be autistic.24 A Danish study found the parent-age link. Grandparent age showed a mixed pattern.25
The parental-age association rests on observational studies that cannot separate age from the many other things that differ between older and younger parents. The grandparental finding is newer and less settled.
An analysis of 12.6 million people across three countries found that older paternal age at the birth of a daughter was linked to autism in her children, while older paternal age at the birth of a son was not, and young maternal age was also linked.26 This disagrees with the Swedish finding above, which reported a grandpaternal effect through both lines, so the split by the middle generation's sex is contested rather than settled.
These are studies of patterns, not proof. Older parents differ from younger parents in many ways.
An analysis of 12.6 million people across three countries found that older paternal age at the birth of a daughter was linked to autism in her children, while older paternal age at the birth of a son was not, and young maternal age was also linked.26 This disagrees with the Swedish finding above, which reported a grandpaternal effect through both lines, so the split by the middle generation's sex is contested rather than settled.
Recurrence in families Established
If one child is autistic, a later sibling is more likely to be autistic too. In a study that followed 664 infant siblings from early life, 18.7% were later diagnosed.14 A 2024 update of the same network, with 1,605 infants, found 20.2%, not significantly higher than before.27 These cohorts are enriched for families who already have an autistic child, so their figure runs higher than in whole-population registers. The risk is higher when the autistic child is a girl, when the infant is a boy, and when more than one family member is autistic; race and maternal education were also linked, pointing to social as well as genetic factors.27
Whole-population registers put the absolute recurrence lower. In Sweden, a full sibling of an autistic child had about 10 times the risk of the general population, a half-sibling about 3 times, and a cousin about 2 times.2 A Danish cohort found an overall relative recurrence risk of about 7, unchanged over time, with the risk higher for maternal half-siblings than paternal ones, which its authors read as pointing to pregnancy and intrauterine factors as well as genes.28 A six-site study found an 8.4-fold increase after an older sibling with autism, a 17.4-fold increase for childhood autism, and about a 2-fold increase for cousins.29
No study found as of October 2026 has measured directly how often the children of autistic parents are autistic. The closest evidence is the sibling and family data above.
If one child is autistic, a later brother or sister is more likely to be autistic too. In one study, about 19 in 100 younger siblings were autistic.14 A later study of the same group found about 20 in 100.27 These families had volunteered, so the number is higher than in whole-country studies. The chance is higher when the autistic child is a girl. It is also higher when the new baby is a boy, or when more than one family member is autistic.27
Whole-country studies find a bigger relative risk but a lower overall chance. In Sweden, a brother or sister had about 10 times the usual risk. A half-sibling had about 3 times. A cousin had about 2 times.2 A Danish study found a risk of about 7 times.28
No study found as of October 2026 has measured how often the children of autistic parents are autistic. The closest evidence is the family studies above.
Prospective infant-sibling cohorts and whole-population registers agree that a sibling's risk is raised. The prospective cohorts are enriched for affected families and give a higher absolute figure than registers; the two prospective figures come from one research network. No study found as of October 2026 has measured directly how often the children of autistic parents are autistic.
Many big studies agree on this. No study found as of October 2026 has measured how often the children of autistic parents are autistic.
In the community Community
Many autistic writers describe autism as running in their families, and describe noticing it in themselves after a child was diagnosed. Jules Edwards, an autistic parent, writes that after her children were diagnosed she looked for the cause and found it in "heritable genetics", and began to recognize autistic traits in herself.c1 The editors of Sincerely, Your Autistic Child write that "many parents of autistic kids were actually autistic, but undiagnosed", and that several of their authors are autistic parents of autistic children.c2 Orion Kelly writes that "if you are a parent of an Autistic child or a grandparent of an Autistic child, chances are one of you are also Autistic."c3
Many autistic writers say autism runs in their families. Some noticed it in themselves after a child was diagnosed. Jules Edwards is an autistic parent. She writes that after her children were diagnosed, she looked for the cause. She found it in "heritable genetics", and saw autistic traits in herself.c1 The editors of Sincerely, Your Autistic Child write that many parents of autistic children were autistic too. They had just never been diagnosed.c2 Orion Kelly writes that a parent or grandparent of an autistic child is often autistic too.c3
How autistic writers describe autism in their families. These are accounts of lived experience, not research findings.
This is how autistic writers describe autism in their families. These are their own experiences, not research findings.
Sources
Research
- 1Visscher PM, Hill WG, Wray NR. Heritability in the genomics era: concepts and misconceptions. Nature Reviews Genetics, 2008. https://doi.org/10.1038/nrg2322
- 2Sandin S, Lichtenstein P, Kuja-Halkola R, Larsson H, Hultman C, Reichenberg A. The Familial Risk of Autism. JAMA, 2014. https://doi.org/10.1001/jama.2014.4144
- 3Tick B, Bolton P, Happé F, Rutter M, Rijsdijk F. Heritability of autism spectrum disorders: a meta-analysis of twin studies. Journal of Child Psychology and Psychiatry, 2016. https://doi.org/10.1111/jcpp.12499
- 4Colvert E, Tick B, McEwen F, et al. Heritability of Autism Spectrum Disorder in a UK Population-Based Twin Sample. JAMA Psychiatry, 2015. https://doi.org/10.1001/jamapsychiatry.2014.3028
- 5Bai D, Yip BHK, Windham GC, et al. Association of Genetic and Environmental Factors With Autism in a 5-Country Cohort. JAMA Psychiatry, 2019. https://doi.org/10.1001/jamapsychiatry.2019.1411
- 6Hallmayer J, Cleveland S, Torres A, et al. Genetic Heritability and Shared Environmental Factors Among Twin Pairs With Autism. Archives of General Psychiatry, 2011. https://doi.org/10.1001/archgenpsychiatry.2011.76
- 7Gaugler T, Klei L, Sanders SJ, et al. Most genetic risk for autism resides with common variation. Nature Genetics, 2014. https://doi.org/10.1038/ng.3039
- 8Grove J, Ripke S, Als TD, et al. Identification of common genetic risk variants for autism spectrum disorder. Nature Genetics, 2019. https://doi.org/10.1038/s41588-019-0344-8
- 9De Rubeis S, He X, Goldberg AP, et al. Synaptic, transcriptional and chromatin genes disrupted in autism. Nature, 2014. https://doi.org/10.1038/nature13772
- 10Fu JM, Satterstrom FK, Peng M, et al. Rare coding variation provides insight into the genetic architecture and phenotypic context of autism. Nature Genetics, 2022. https://doi.org/10.1038/s41588-022-01104-0
- 11de Wit MM, Morgan MJ, Libedinsky I, et al. Systematic Review and Meta-Analysis: Phenotypic Correlates of the Autism Polygenic Score. JAACAP Open, 2025. https://doi.org/10.1016/j.jaacop.2025.04.001
- 12Turner TN, Coe BP, Dickel DE, et al. Genomic Patterns of De Novo Mutation in Simplex Autism. Cell, 2017. https://doi.org/10.1016/j.cell.2017.08.047
- 13More RP, Warrier V, Brunel H, et al. Identifying rare genetic variants in 21 highly multiplex autism families: the role of diagnosis and autistic traits. Molecular Psychiatry, 2023. https://doi.org/10.1038/s41380-022-01938-4
- 14Ozonoff S, Young GS, Carter A, et al. Recurrence Risk for Autism Spectrum Disorders: A Baby Siblings Research Consortium Study. Pediatrics, 2011. https://doi.org/10.1542/peds.2010-2825
- 15Tammimies K, Marshall CR, Walker S, et al. Molecular Diagnostic Yield of Chromosomal Microarray Analysis and Whole-Exome Sequencing in Children With Autism Spectrum Disorder. JAMA, 2015. https://doi.org/10.1001/jama.2015.10078
- 16Harris HK, Sideridis GD, Barbaresi WJ, Harstad E. Pathogenic Yield of Genetic Testing in Autism Spectrum Disorder. Pediatrics, 2020. https://doi.org/10.1542/peds.2019-3211
- 17Sucksmith E, Roth I, Hoekstra RA. Autistic Traits Below the Clinical Threshold: Re-examining the Broader Autism Phenotype in the 21st Century. Neuropsychology Review, 2011. https://doi.org/10.1007/s11065-011-9183-9
- 18Rubenstein E, Chawla D. Broader Autism Phenotype in Parents of Children with Autism: A Systematic Review of Percentage Estimates. Journal of Child and Family Studies, 2018. https://doi.org/10.1007/s10826-018-1026-3
- 19Robinson EB, Lichtenstein P, Anckarsäter H, Happé F, Ronald A. Examining and interpreting the female protective effect against autistic behavior. PNAS, 2013. https://doi.org/10.1073/pnas.1211070110
- 20Wigdor EM, Weiner DJ, Grove J, et al. The female protective effect against autism spectrum disorder. Cell Genomics, 2022. https://doi.org/10.1016/j.xgen.2022.100134
- 21Sandin S, Yip BHK, Yin W, et al. Examining Sex Differences in Autism Heritability. JAMA Psychiatry, 2024. https://doi.org/10.1001/jamapsychiatry.2024.0525
- 22Dougherty JD, Marrus N, Maloney SE, et al. Can the "female protective effect" liability threshold model explain sex differences in autism spectrum disorder? Neuron, 2022. https://doi.org/10.1016/j.neuron.2022.06.020
- 23Wu S, Wu F, Ding Y, Hou J, Bi J, Zhang Z. Advanced parental age and autism risk in children: a systematic review and meta-analysis. Acta Psychiatrica Scandinavica, 2017. https://doi.org/10.1111/acps.12666
- 24Frans EM, Sandin S, Reichenberg A, et al. Autism Risk Across Generations: A Population-Based Study of Advancing Grandpaternal and Paternal Age. JAMA Psychiatry, 2013. https://doi.org/10.1001/jamapsychiatry.2013.1180
- 25Gao Y, Yu Y, Xiao J, et al. Association of Grandparental and Parental Age at Childbirth With Autism Spectrum Disorder in Children. JAMA Network Open, 2020. https://doi.org/10.1001/jamanetworkopen.2020.2868
- 26Reichenberg A, Schendel D, Gissler M, et al. Risk for Autism Across Generations. Biological Psychiatry, 2026. https://doi.org/10.1016/j.biopsych.2026.04.023
- 27Ozonoff S, Young GS, Bradshaw J, et al. Familial Recurrence of Autism: Updates From the Baby Siblings Research Consortium. Pediatrics, 2024. https://doi.org/10.1542/peds.2023-065297
- 28Grønborg TK, Schendel DE, Parner ET. Recurrence of Autism Spectrum Disorders in Full- and Half-Siblings and Trends Over Time: A Population-Based Cohort Study. JAMA Pediatrics, 2013. https://doi.org/10.1001/jamapediatrics.2013.2259
- 29Hansen SN, Schendel DE, Francis RW, et al. Recurrence Risk of Autism in Siblings and Cousins: A Multinational, Population-Based Study. Journal of the American Academy of Child and Adolescent Psychiatry, 2019. https://doi.org/10.1016/j.jaac.2018.11.017
Community sources
- c1Ashburn M, Edwards J, Onaiwu M. I Will Die On This Hill: Autistic Adults, Autism Parents, and the Children Who Deserve a Better World. Jessica Kingsley Publishers, 2023. Chapter "Identification".
- c2Ballou EP, daVanport S, Onaiwu MG, eds. Sincerely, Your Autistic Child: What People on the Autism Spectrum Wish Their Parents Knew About Growing Up, Acceptance, and Identity. Beacon Press, 2021. Letter from the Editors.
- c3Kelly O. Autism Feels ...: An Earthling's Guide. Dean Publishing, 2023. Chapter 3, "Autism Facts".