“Autism is genetic.”

It is a powerful headline—but by itself, it is not a complete explanation.

On 15 September 2026, the National Autistic Society (NAS), UK announced that its research team had reviewed 83 research papers concerning the causes of autism and had updated its public position to state that autism has a genetic basis. NAS also states that there is no single “autism gene,” that other factors can contribute in less common circumstances, and that there is currently no genetic test that diagnoses autism or predicts with certainty whether an individual child will be autistic.

The announcement has understandably generated discussion.

Some of that discussion is about genetics itself. Some is about how genetic evidence should be communicated. And some is about an even more important question:

What should our growing understanding of autism biology change about the way we support autistic people and families?

This article looks at that question in detail.
The first distinction: genetic does not mean deterministic

When people hear that a condition is “genetic,” it is easy to interpret that as:
A gene causes it → therefore the outcome is predetermined.

That is not how the genetics of autism works.
Autism has a 'complex genetic architecture', involving many forms of genetic variation rather than one single causal gene. Research has identified contributions from both common genetic variation and rare variants, including inherited variants and variants that arise de novo.

So the scientifically useful statement is not:
“One gene causes autism.”
It is closer to:

Genetic variation makes a substantial contribution to the likelihood and biological development of autism, but that contribution operates through a complex set of genetic pathways.
That difference matters enormously.

What does “80% heritability” actually mean?

One of the most frequently discussed findings in autism genetics comes from a large five-country population study published in 'JAMA Psychiatry'.

The study analysed 2,001,631 people, including 22,156 people diagnosed with autism spectrum disorder, across Denmark, Finland, Sweden, Israel and Western Australia. The median country-specific estimate of ASD heritability was 80.8%, although the estimates varied substantially between countries—from 50.9% in Finland to 86.8% in Israel.

That sounds enormous.
But there is a critical statistical point:

-Heritability is a population-level measure.

It does not mean:

• 80% of an individual's autism is caused by genes.
• 80% of a child's clinical presentation can be explained by a genetic test.
• 80% of autism is caused by one set of genes.
• An individual's autism can be predicted with 80% certainty.

Instead, heritability estimates describe how much of the 'variation in a trait within a particular population and study context can be statistically attributed to genetic differences between individuals'.

This distinction between 'population-level variance' and 'individual-level causation' is fundamental.

A statistic about a population cannot, by itself, tell us why a particular child is autistic.

And this is one of the reasons autism genetics should be communicated with considerably more precision than a headline allows.

Autism does not have a single “autism gene”

The current evidence points toward a much more complicated biological picture.
Large genetic studies have identified numerous autism-associated genes and variants.

For example, one Nature Genetics study analysed coding variation in 63,237 individuals and identified 72 genes associated with ASD at a stringent statistical threshold, with substantially more implicated when a less stringent threshold was used. The evidence included de novo protein-truncating variants, damaging missense variants and copy-number variants.

Another large analysis incorporating 42,607 autism cases identified additional moderate-risk genes by examining both rare inherited and de novo coding variants.

This tells us something important:
Autism is not one genetic condition with one molecular explanation.

Rather, multiple genetic pathways can contribute to neurodevelopmental differences.
This complexity also helps explain why autistic people can have very different clinical and developmental profiles.

Common variants, rare variants and de novo variants
Part of the complexity comes from the different kinds of genetic variation involved.

-Common genetic variation

Common variants are genetic differences that occur relatively frequently in a population. Many such variants may each contribute a small amount to autism susceptibility.

Collectively, these effects can contribute to polygenic risk.

Rare variants

Rare variants may have larger effects in some individuals.

Large sequencing studies have identified rare variants and copy-number changes that are associated with autism and developmental differences. (Nature)

De novo variants

Some variants are not inherited from either parent but arise during the formation of reproductive cells or early development.

Large autism sequencing studies have found evidence for de novo variants among the genetic factors associated with autism. (Nature)

These mechanisms do not operate as mutually exclusive boxes.

An individual's neurodevelopmental profile can reflect a combination of different genetic influences.

---

Why autism heterogeneity matters so much

This may be the most clinically important part of the discussion.

Autism is highly heterogeneous.

Autistic people can differ considerably in:

• communication
• social interaction
• sensory processing
• repetitive behaviours and interests
• adaptive functioning
• motor development
• cognitive profile
• language
• emotional regulation
• co-occurring conditions
• functional participation

Research increasingly suggests that this heterogeneity is not merely a descriptive clinical problem—it also exists at the genetic and molecular level.

A 2025 Nature Genetics study analysed broad phenotypic and genetic data from 5,392 autistic individuals and identified clinically meaningful classes associated with different genetic and molecular programs. The authors concluded that patterns of phenotype and clinical outcome corresponded to different combinations of common, de novo and inherited genetic variation. (Nature)

Other research has similarly identified relationships between different genetic risk profiles and different phenotypic features within autism.

This creates an important clinical warning:

We should be very careful about moving from “genetic contribution” to “genetic explanation of an individual child.”
The two are not equivalent.

Does this mean environmental factors don't matter?
No.

This is another area where the conversation can become unnecessarily binary.
Genetics and environment should not automatically be treated as competing explanations.

The five-country JAMA Psychiatry study found substantial genetic contributions while also estimating other components of variation. The authors' models included genetic, maternal and environmental effects and reported differences across countries.

Other genetic studies have similarly described autism risk as multifactorial, involving rare variants, polygenic risk and other influences.

The more accurate question is therefore not:

“Is autism genetic or environmental?”

It is:

“How do genetic, developmental and environmental factors interact to influence neurodevelopment and clinical presentation?”
That is a much harder scientific question—and one that remains active.

Can a genetic test diagnose autism?

Currently, no.

The National Autistic Society explicitly states that there is no genetic test that can diagnose autism, and no genetic test that can predict with certainty whether a child will be autistic.
This distinction is important because genetic testing can still have a role in clinical practice in selected cases.

For example, a clinician may investigate a possible genetic or syndromic condition when a child's developmental profile or other clinical findings suggest that such testing may be appropriate.

But that is different from having a genetic test that simply answers:

“Is this child autistic?”
We do not currently have such a definitive genetic test.

So what does this mean for parents?

Perhaps the biggest practical issue is communication.

When a parent hears:
“Autism is genetic.”
they may reasonably ask:

• What does this mean for my child?
• Did I pass this on?
• Could I have prevented it?
• Will my next child be autistic?
• Can a test predict the future?
• Is there a treatment for the genetic cause?

The current evidence does not support simple answers to those questions.
Genetic contribution should 'not' be translated into parental blame.

Nor should it be turned into promises that genetic testing can predict an individual child's future.
Science becomes harmful when a nuanced finding is converted into a simplistic message.

Genetics does not tell us what support a child needs

This is where the conversation moves from 'etiology to clinical care'.

A genetic finding does not tell us whether a particular child needs help with:
-communication.
-sensory processing.
-motor development.
-self-care.
-emotional regulation.
-social participation.
-learning.
-school readiness.
-adaptive functioning.

Those questions require understanding the child.
That means assessment of strengths, challenges, context, participation and functional goals.
Two autistic children may have entirely different priorities despite sharing the same diagnostic label.

And therefore:

A diagnosis is not a therapy plan.
This is where rehabilitation has an important role
Autism support is not simply about finding one therapy that “treats autism.”
The clinically meaningful question is:
What does this individual need support with?
That distinction changes how we think about intervention.

For one child, support may prioritise communication.
For another, participation in school.
For another, sensory-motor development.
For another, activities of daily living.
For another, emotional or behavioural regulation.
For another, social participation and functional independence.

Depending on the needs of the person and family, appropriate professionals may include occupational therapists, speech and language professionals, psychologists, developmental paediatricians, educators, behaviour professionals and others.

The important principle is individualisation and coordination, not simply increasing the number of therapies.

We should be careful with the word “treatment”

Autism is a neurodevelopmental condition, and autistic people can have very different support needs across their lives.
That makes “treatment” a complicated word.
There is no single intervention that explains or resolves autism as a whole.
Instead, the focus of support should be on meaningful outcomes for the individual:
-communication
-participation
-learning
-independence
-quality of life
-family functioning
-wellbeing

and the ability to participate meaningfully in home, school, community and later adult life.
This distinction is particularly important because families can encounter programs promoted as “breakthrough therapies” or universal solutions.

#More intervention is not automatically better intervention.
The relevant questions are:
-What is the goal?
-What evidence supports the intervention?
-For whom does it work?
-What outcome are we measuring?
-What are the potential benefits and harms?
-How will we know whether it is actually helping this child?

Understanding biology should improve intervention research

This is where I believe the genetics conversation becomes genuinely exciting.
The value of genetics is not only in explaining why autism occurs.
It may eventually help researchers understand:
-biological pathways
-developmental mechanisms
-subgroups
-co-occurring conditions
-treatment response
-potential therapeutic targets

The 2025 Nature Genetics research is particularly interesting because it attempts to connect different patterns of phenotype with different genetic and molecular programs rather than treating autism as biologically uniform.

That direction could eventually contribute to more biologically informed research.
But we are not there yet.

It would be premature to take genetic associations and assume they immediately translate into individually tailored treatments.
That translation requires much more research.

From “Which therapy?” to “Which outcome?”

This may be one of the biggest changes the field needs.
Instead of asking:

“Which therapy is best for autism?”

we should ask:

“Which intervention produces which outcome, for which child, under which circumstances?”

That is a much more scientifically useful question.
For example, if a child has difficulty participating in school, we should not automatically measure success through symptom reduction alone.

We should also ask:
Can the child participate more effectively?
Can they communicate their needs?
Can they manage daily routines?
Can they engage in meaningful activities?
Can the family support them more effectively?
Has their quality of life improved?
These are the outcomes that connect research to real life.

Why “miracle therapy” narratives are particularly dangerous

When families are searching for answers, they are vulnerable to certainty.
A new program can sound compelling when it promises:
“Complete recovery.”
“Normalisation.”
“Cure.”
“One therapy for every child.”

But complex neurodevelopmental conditions do not lend themselves easily to universal solutions.
A responsible clinical culture should therefore reward:

-evidence over marketing
-individual goals over generic promises
-measured outcomes over testimonials alone
-professional collaboration over isolated approaches
and long-term wellbeing over short-term claims.

That does not mean innovation should stop.
It means innovation should be tested.

A second important issue: population evidence versus individual care

This is where the criticism I received after my earlier discussion became particularly valuable.
Genetic studies are generally designed to answer questions about populations.

Clinical care is delivered to individuals.
Those are different levels of inference.

A study can identify a statistically meaningful association across thousands or millions of people without allowing us to conclude that a specific genetic factor caused one child's presentation.

This is not a weakness unique to autism research.
It is a fundamental principle of interpreting biomedical evidence.
And it is why good clinicians need both:

population evidence
and
individual clinical reasoning.

What the NAS review changes—and what it doesn't

The September 2026 NAS announcement is important primarily as a public evidence-position update.

NAS says its research team reviewed 83 papers and now explicitly states that autism has a genetic basis.
But it is important not to confuse that announcement with the emergence of a completely new scientific discovery in September 2026.

The broader scientific literature had already established substantial evidence for genetic involvement in autism years earlier, including large-scale genetic studies and population studies. (JAMA Network)

So the most accurate way to describe the development is:

The National Autistic Society has updated its public position after reviewing 83 research papers; this sits within a much larger and evolving scientific literature on the genetic architecture of autism.

That distinction matters because scientific communication should be precise.

What we still don't know

Despite major progress, important questions remain.

We still don't fully understand:
Why particular genetic combinations lead to different developmental trajectories?
How genetic and environmental factors interact across development?
Why some genetic variants are associated with very different phenotypes in different people?
How biological subgroups should translate into clinical practice?
Which interventions work best for which individuals and outcomes?
How to identify meaningful biomarkers without reducing a person to a biological profile?
These are enormous research questions.

And they require collaboration between genetics, neuroscience, developmental medicine, psychology, education, rehabilitation and lived-experience communities.

Where I believe the next conversation should go

My interest in this subject is increasingly moving beyond:

“What causes autism?”
toward:
“How can our growing understanding of autism biology help us provide better support?”
I am actively exploring the intervention and treatment evidence around autism—including what is well supported, where evidence is weak or inconsistent, how outcomes should be measured, and where significant research gaps remain.

I am particularly interested in the intersection of:
neurodevelopment × rehabilitation × clinical evidence × AI

because the next generation of healthcare systems should not simply collect information.
They should help clinicians navigate evidence, individualise care and measure outcomes more intelligently.

But that work needs to be grounded in science.
Not hype.
Not miracle claims.
And not one-size-fits-all solutions.

A more responsible summary

So, what should we take away from the current evidence?
Autism has a substantial genetic contribution.

That is supported by large population and genetic studies.
'Autism is genetically complex.'

Multiple types of genetic variation contribute, and there is no single “autism gene.”
Heritability is not individual causation.

A population statistic cannot explain one child's presentation.
There is currently no genetic test that diagnoses autism or predicts it with certainty.
Autism is highly heterogeneous.

Different people can have substantially different developmental and clinical profiles, and genetic research is beginning to map some of that heterogeneity.
Understanding biology does not remove the need for therapy and support.
It makes the need for evidence-informed, individualised support even more important.
And perhaps the most important principle is this:

Understand the biology.
Understand the individual.
Measure what matters.
Provide support that has evidence behind it.

The purpose of better science should never be to reduce a child to a diagnosis, a statistic or a genetic profile.

It should help us understand them better.

For parents and caregivers

If your child is autistic or is being evaluated for autism, a genetics headline should not be interpreted as a complete explanation of your child's future.

Genetic research is an important part of understanding autism, but everyday support decisions remain individual and should be discussed with appropriately qualified professionals.

A child is more than their diagnosis.
Strengths, interests, communication, relationships, participation and quality of life matter too.

#For clinicians and therapists
The genetics conversation is a reminder that the field is moving toward greater biological complexity.
But clinical practice should remain grounded in:
-individual assessment
-functional goals
-evidence-informed intervention
-multidisciplinary collaboration
-outcome measurement
and continuous reassessment.

The goal is not to find more therapies to add to a child's schedule.
The goal is to find the right support for the right need at the right time.

For researchers and innovators

There is an enormous opportunity ahead—but also a responsibility.
The future of autism research should connect:
genomics → neuroscience → phenotype → intervention → outcomes

without skipping the difficult middle steps.
That means validating biomarkers.
Understanding heterogeneity.
Testing interventions rigorously.
Measuring meaningful outcomes.
And ensuring that technological advances actually improve the lives of autistic people and families.\

Final thought

The phrase “Autism is genetic”may be useful as a starting point.
It should not be the end of the conversation.
The more we learn about autism, the more obvious one thing becomes:

There is no single story of autism.

There are individual people with different biology, different strengths, different challenges, different environments and different goals.

And the responsibility of healthcare is to respond to that complexity—not erase it.

Sources & further reading

National Autistic Society (UK). Autism is genetic, published 15 September 2026. NAS reports a review of 83 research papers and its updated position on genetic contribution.

Bai D, et al. Association of Genetic and Environmental Factors With Autism in a 5-Country Cohort. JAMA Psychiatry. 2019;76(10):1035–1043. The study included 2,001,631 individuals and estimated median ASD heritability of 80.8%, with variation across countries.

Satterstrom FK, et al. Large-scale exome sequencing study implicates both developmental and functional pathways in autism. Related large-scale genetic analyses have identified many autism-associated genes and both inherited and de novo variation.

Fu JM, et al. Rare coding variation provides insight into the genetic architecture and phenotypic context of autism. Nature Genetics. 2022;54:1320–1331.

Litman A, et al. Decomposition of phenotypic heterogeneity in autism reveals underlying genetic programs. Nature Genetics. 2025;57:1611–1619.

Leppa VM, et al. Genetic correlates of phenotypic heterogeneity in autism. Nature Genetics. 2022;54:1293–1304.

Levy D, et al. Research on combined rare and polygenic genetic influences in autism and their relationship to phenotypic variation.

This article is an educational review and is not a substitute for individualized medical, genetic or therapeutic advice.