Autism, diagnosis and family

Why does autism seem to run in my family?

Is autism genetic? A careful explanation of heritability, family patterns, genetic variation and epigenetics—without turning family resemblance into somebody else’s diagnosis.

Last reviewed: 19 September 2026

After an autism diagnosis, some people look around their family and recognise familiar ways of moving through the world: strong routines, sensory sensitivities, a longstanding intense interest, a particular social style. Others look around and recognise nobody at all.

Neither response tells you the whole genetic story. Autism can cluster in families without there being an identifiable autistic parent, sibling or grandparent. And not seeing a pattern does not make your diagnosis unusual, less real or in need of a hidden family explanation.

Is autism genetic? The short answer

Genetic differences make a substantial contribution to autism. But autism usually does not follow a simple single-gene inheritance pattern, where one identifiable gene is passed from one parent and reliably produces one outcome.

“Runs in families” does not necessarily mean “is passed down in a simple way”. For many people, autism is better understood as the result of many genetic and developmental influences coming together. Some influences are inherited; some can be new genetic changes; many are not identifiable in one person with current testing.[15]

That is not an evasive answer. It is the current scientific picture: strong evidence for a substantial genetic contribution, alongside real uncertainty about the particular route for any one autistic person.

What does “highly heritable” actually mean?

Large twin and family studies estimate that genetic differences account for a substantial share of the variation in likelihood of an autism diagnosis within the populations they studied. A twin-study meta-analysis reported heritability estimates ranging from 64% to 91%. A large Swedish family study estimated 83%, while a five-country register study reported a median estimate of 80.8%, with estimates differing between countries and models.[8][9][10]

Those figures are easy to misunderstand. An estimate of 80% does not mean that 80% of one person’s autism was caused by genes and 20% by “environment”. It is a population statistic: under a particular study design and set of assumptions, it describes how much variation in liability between people is statistically attributable to genetic differences in that population.

It cannot divide an individual life into genetic and environmental portions. It does not tell us which genes matter for a person, whether a parent passed autism on, or what someone’s personality, support needs or future will be. It also does not mean that the unexplained remainder is a known list of parental actions or environmental causes. Heritability changes with the population, diagnostic practice, follow-up, methods and assumptions used.[8][9]

There is not one autism gene

Autism’s genetic architecture is varied. Many common DNA variants each have very small average effects. Taken together, they can make a meaningful polygenic contribution; individually, they are not diagnostic “autism genes”. A large 2019 genome-wide association study identified common associated loci in a much larger autism sample; that supports a polygenic contribution, not a DNA test that can determine whether an individual is autistic.[17]

Rarer variants can sometimes have larger effects. These can be inherited, or de novo: present in a person but not detected in either biological parent’s tested DNA, having arisen in an egg or sperm cell, or during early development. Copy-number variants are another kind of change, involving a missing or extra stretch of DNA. Large sequencing studies find contributions from rare inherited, de novo and structural variation as well as common variation.[4][15]

The important practical point is not a catalogue of genes. Two autistic people may reach an autism diagnosis through quite different combinations of influences. The reverse is also important: relatives can share some of the same genetic influences without having the same developmental outcome. Even when a rare variant is clinically important, it may have incomplete penetrance or variable expression: carrying it does not make one person’s outcome a template for another’s.[4][16]

Why can autism seem to run through a family?

Family studies consistently find that autism diagnoses are more common among close biological relatives of autistic people than in comparison groups. That pattern is called familial aggregation. It is compatible with inherited genetic contributions, but it does not identify one causal parent or turn family resemblance into diagnosis.

In a large prospective study of 1,605 later-born biological infant siblings of an autistic child, 20.2% received an autism diagnosis by about age three. This is sibling-recurrence evidence for a particular high-familial-risk research cohort: it is not a prediction for every sibling, adult relative or future child.[1]

Research also uses the term broader autism phenotype for sub-diagnostic autism-related traits in some relatives. It is a research construct, not “partly autistic” and not a diagnosis. A systematic review found that estimates varied enormously between studies because they used different measures, samples and thresholds. A family member can have routines, sensory preferences, a social style or an intense interest for many reasons. Recognising a resemblance is not the same as diagnosing somebody.[12]

A missing past diagnosis in an older relative settles very little. A diagnosis that was never sought or offered cannot reliably answer a present question about resemblance. It does not establish that the person was autistic, and it does not establish that they were not.

You can notice a family resemblance without needing to turn it into somebody else’s diagnosis.

What if nobody else in my family seems autistic?

This deserves its own answer: nothing is wrong with that picture.

You may be the only diagnosed autistic person in a large family. You may have limited biological-family information, have been adopted, be estranged from relatives, or have no reliable developmental history to compare. Relatives may share no obvious traits, share some traits without meeting diagnostic criteria, or have had different lives and different access to assessment.

You may also have a combination of common inherited variants, rare inherited variants, de novo variation, other developmental influences, or influences that current science cannot identify individually. Looking at relatives cannot tell you which explanation applies. A de novo variant is one possible route, not a catch-all explanation for every apparently isolated autism.[4][15]

A family appearance is therefore not a genetic test. It cannot prove where your autism came from, and it cannot disprove it.

Does genetics mean a sibling or child will be autistic?

Genetics can increase group-level likelihood without giving a simple personal forecast.

The 20.2% figure above is often repeated without its setting. It concerned later-born infant siblings of an autistic child, recruited to a prospective international research study and assessed in early childhood. It should not be repurposed as “the chance” for an autistic adult’s child.[1]

A Swedish register study separately found higher group-level odds of autism diagnoses among children with a parent recorded as autistic. It reported relative odds, not a universally applicable absolute percentage. It cannot identify whether a mother or father “caused” autism, and it cannot tell one family what will happen.[11]

This guide is not a reproductive decision tool. Autism is not an outcome to be evaluated from outside, and population studies do not tell us the value, needs or life of a particular child.

Is autism inherited from the mother or father?

There is no evidence-based shortcut that assigns autism to one parent. Common genetic variants contributing to autism liability are generally inherited across the genome from both biological parents, while rarer variants can be inherited in different ways or arise de novo. A large multinational family study found no support for a separate maternal-effect contribution in its model; that is not a test of whether a particular person’s mother or father “gave them autism”.[9]

In an individual family, DNA is only one part of an incomplete developmental account. Family stories, resemblance and a consumer ancestry result cannot identify a responsible parent. It is more accurate to say that inherited and new genetic variation can both contribute than to look for a maternal or paternal source.

What about epigenetics?

Epigenetics describes ways cells regulate how genetic information is used without changing the underlying DNA sequence. DNA methylation, chromatin and histone changes are among the mechanisms researchers study. They are ordinary parts of development: a neuron and a blood cell use the same DNA differently because they need different gene programmes.

This makes epigenetics relevant to autism research, especially because some genetic variants associated with autism affect gene-regulation machinery. It does not mean that an experience simply switches “autism genes” on or off.[4][6]

Researchers have reported DNA-methylation differences associated with autism in brain tissue, placenta, cord blood, blood and other samples. These findings are heterogeneous. A blood finding is not automatically a finding in a fetal brain, and large blood-based studies have not produced a validated individual autism test.[6][7]

Most studies can identify association, not direction. A difference may be part of a causal pathway; it may also be a consequence of other biological or developmental differences; a correlated process; a difference in cell composition; a technical artefact; or partly downstream of inherited DNA variation that itself affects regulation. Well-designed prospective, genetically informed and experimental work can test narrower causal questions, but the general causal account remains under investigation.[6][7]

That is why epigenetics must not be used to revive parental blame in modern language. It does not show that a parent’s emotions, ordinary diet, family stress, pregnancy experience or care “caused autism by switching genes on”. It also does not establish transgenerational epigenetic inheritance of autism in humans. Animal experiments can test mechanisms across generations; they are not proof of an inherited autism mechanism in human families.[7][14]

Genes and environments are therefore not opposing teams. Genetic variation can shape biological pathways and responses; developmental conditions can matter; epigenetic regulation is one way cells organise biological activity. Complexity is not evidence that any particular exposure explains one person’s autism.

Should autistic adults have genetic testing?

Current adult autism diagnosis is clinical and developmental. NICE says not to use biological tests, genetic tests or neuroimaging routinely for diagnosis as part of a comprehensive adult autism assessment.[2]

That does not mean genetics is never relevant. NICE says that, on an individual basis and with clinical judgement, further genetic investigation may be considered where there are particular dysmorphic features, congenital anomalies and/or evidence of a learning disability, following regional genetics advice.[2] The question in that situation is usually whether there may be a wider genetic or chromosomal condition relevant to somebody’s health or developmental presentation, not whether a DNA test can prove ordinary adult autism.

A consumer DNA test cannot establish whether you are autistic. Equally, not finding an identified variant does not make an autism diagnosis less valid. The science of why someone is autistic and the clinical question of whether they meet diagnostic criteria are different questions.

What can I conclude about my own family?

Genetics can explain why family patterns occur without giving any one person a complete family tree of causation.

You may find a new framework makes some similarities easier to notice. You may not. You do not need to identify an autistic relative, find a genetic variant, or settle an inheritance story for your diagnosis to be real.

The most accurate conclusion is usually modest: autism can be strongly familial and genetically influenced, while each person’s developmental route remains individual and often partly unknown.