Autism genetic testing: what it finds, what it costs, and what to ask

What autism genetic testing can and cannot do

If genetic testing has just been offered to you, the most useful thing to be clear about is what is actually on the table.

Autism genetic testing is not a test for autism. It is a search for a genetic condition underneath a diagnosis your child already has. That diagnosis was made clinically — from history, observation, and how your child communicates, plays and relates to people. The American Academy of Pediatrics clinical report on identifying, evaluating and managing children with autism spectrum disorder expects primary care clinicians to know the diagnostic criteria for autism and, as a separate matter, appropriate etiologic evaluation — the search for a cause (Hyman 2020). Autism is diagnosed by evaluation, not by a laboratory. If you are still working toward that diagnosis, start with autism testing explained.

So no sample of blood or saliva puts a child on the spectrum, and none takes a child off it. If a clinic or a company implies otherwise, that is a reason to be careful with everything else they tell you.

What the test can do is find a named genetic condition in a minority of children. Pooled across the autism cohorts in a systematic review of 103 clinical sequencing studies, the diagnostic yield in autism was 17.1% (Stefanski 2021). Read that number honestly in both directions: roughly one child in six gets an explanation, and roughly five in six do not.

What “a cause” actually means

When a lab reports a genetic finding, it is saying that a change in a particular gene or a particular stretch of chromosome is enough to explain part of your child’s developmental picture. It is not saying that this change is your child.

Nearly everything known about these conditions is population-level information — what has been observed across the group of people described with that finding so far, often a small group, often studied because they were the most affected. It is a description of a range, not a prediction about one person.

That matters most in the appointment. A named condition may come with a screening schedule, a specialist to see, a research registry to join and a parent community that understands your week. It does not come with a ceiling. If anyone tells you what your child will never do on the basis of a gene name, they have gone beyond what the science supports.

Three more things testing will not do:

• It will not find something in every child. Most results come back with nothing to report.
• It will not replace educational evaluation, therapy decisions or support at school.
• It will not change who your child is. The result was already true before the sample was taken.

And one thing this site will not do: we do not support or promote prenatal or embryo screening for autism. Testing a child who already has a diagnosis, so that their medical care can be better, is a different act from screening a pregnancy or an embryo to decide whether an autistic person should exist. Everything on this page is about the first.

Who is offered testing, and what the guidelines say

Professional bodies have been recommending a genetics evaluation after an autism diagnosis for well over a decade. What has changed is which test they recommend first.

ACMG, 2013. The American College of Medical Genetics and Genomics guideline on identifying the etiology of autism spectrum disorders sets out the ground rules that still hold. Confirm the autism diagnosis before investigating. Discuss the testing options, the diagnostic yields and the family investment involved before proceeding. Coordinate with the child’s regular medical home, and build an individual plan from that child’s history and features rather than running a fixed list (Schaefer 2013). The instruction to talk about yields with the family up front is the one most often skipped, and it is the one you can ask for by name.

AAP, 2020. The American Academy of Pediatrics clinical report says primary care clinicians should be familiar with appropriate etiologic evaluation — the search for a cause — alongside diagnosis and the co-occurring conditions that affect a child’s day (Hyman 2020). In practice this is why your pediatrician can raise genetics without waiting for a specialist to think of it.

Srivastava, 2019. A meta-analysis and multidisciplinary consensus statement compared exome sequencing with chromosomal microarray, which had been the standard first test. Exome yield was 36% overall across the included studies, 31% for isolated neurodevelopmental conditions and 53% where other associated conditions were present, against 15–20% in the earlier microarray literature. The group proposed putting exome sequencing at the beginning of the evaluation for unexplained neurodevelopmental conditions rather than at the end.

ACMG, 2021. An evidence-based clinical guideline then made it formal: the ACMG strongly recommends that exome or genome sequencing be considered as a first- or second-tier test for children with congenital anomalies, developmental delay or intellectual disability, noting it has a higher diagnostic yield than standard genetic testing and may be more cost-effective when ordered early (Manickam 2021).

AAP, 2025. The most recent clinical report, on genetic evaluation of the child with intellectual disability or global developmental delay, describes the practical approach for a general pediatrician. Where the clinical picture points to a specific condition, testing should be chosen by phenotype. Where it does not, an agnostic — hypothesis-free — approach is used, with the choice of test based on diagnostic yield and on practical considerations including test complexity and the impact on management (Rodan 2025).

The joint statement, 3 June 2026. The National Society of Genetic Counselors, the American Society of Human Genetics and the American College of Medical Genetics and Genomics issued a short joint statement on genetics and autism. It puts autism at “approximately 1 in 31 children (3.2%)”, says that “hundreds of genes and chromosomal variations are known to increase the likelihood of autism”, and says that better testing and large collaborative studies “have improved the ability to identify likely genetic causes, which can allow early intervention for associated medical conditions to improve outcomes.” We wrote about what it means for your next appointment in three organisations just agreed on genetic testing.

What this means in the room

Guidelines describe what should be available. They do not describe what every family is actually offered, and uptake lags the guidelines. We are not going to put a number on that gap, because we have not found a figure we can stand behind. But if nobody has mentioned genetics to you, that is not a sign it was considered and ruled out on your behalf. It is usually a sign that nobody got to it.

What you can do about it is smaller and more useful than a statistic: ask. “Has a genetics evaluation been offered to us?” is a fair question at any appointment, and “not yet” is a normal answer that a referral fixes.

The tests, one by one

Clinicians rarely say “a genetic test”. They name one, and the names are not self-explanatory. Here is what each one reads, and — just as important — what it misses.

Chromosomal microarray

A chromosomal microarray scans the chromosomes for stretches of DNA that are missing or present more than once. These are called copy number variants. If the genome is a book, a microarray checks whether any chapters have been torn out or bound in twice.

It has been a standard first-tier test for years, it is widely available, and for many families it is the test insurance authorizes first. What it cannot do is read the sentences. A single changed letter inside a gene is invisible to it. So a microarray that finds nothing narrows the possibilities without ruling out a genetic explanation.

Fragile X testing

Fragile X testing looks for one specific kind of change: a repeat expansion in the FMR1 gene, where a short sequence of DNA letters is copied over and over far more times than usual. It needs its own test, because standard sequencing does not reliably count those repeats. It is the easiest thing on this list to fall through the gaps, because sequencing feels comprehensive and on this one point is not. Ask directly: has fragile X been ordered separately?

Exome sequencing

Whole exome sequencing reads the protein-coding part of all roughly 20,000 human genes — about 1 to 2% of the genome, but the part where most currently interpretable disease-causing changes sit. It reads the letters themselves, which is why it finds the single-letter changes a microarray cannot. It is not a replacement for a microarray: in the one study here where the same children had both, the two tests mostly found different things — only 2 of the 15 children with a diagnosis got one from both tests (Tammimies 2015).

Genome sequencing

Genome sequencing reads nearly all of the DNA, including the long stretches between genes. It catches some variant types exome sequencing misses, and it is increasingly offered where it is available. The 2021 ACMG guideline treats exome and genome sequencing together, recommending either as a first- or second-tier test for congenital anomalies, developmental delay or intellectual disability (Manickam 2021).

Gene panels

A panel sequences a pre-selected list of genes tied to a particular clinical picture — an epilepsy panel, an intellectual disability panel. Panels are quicker to interpret and often cheaper. Their limit is structural: a panel cannot find anything outside its own list. In the pooled meta-analysis, exome sequencing found a cause in 27.2% of cases against 22.6% for panel sequencing, a difference that did not reach statistical significance (Stefanski 2021).

Why the numbers move around so much

You will see very different yield figures for what sounds like the same test. The main reason is not the technology. It is who was tested.

The clearest single comparison comes from a study of 258 children with autism recruited consecutively from a developmental pediatric clinic — that is, a realistic mix rather than a hand-picked one. Every child had a microarray; 95 also had exome sequencing as a trio with both parents. Microarray found a molecular diagnosis in 24 children (9.3%). Exome found one in 8 of the 95 (8.4%). Among the 95 children who had both tests, 15.8% had an identifiable genetic cause. The combined yield was significantly higher in the group with the most complex physical features — scored on major congenital anomalies and minor physical anomalies — than in the group with the fewest (Tammimies 2015).

Now set that beside the exome figures in the section above, and beside the pooled autism figure, which the authors of that review drew from cohorts they themselves describe as highly selected (Stefanski 2021).

Same tests. Different populations. The highest of those figures describes children who already had other medical findings and had reached a genetics service; the lowest describes everyone walking through a developmental clinic door. Neither is wrong. If a lab brochure quotes you one number, the question to ask is who was in that study, and are they like my child?

One thing to be skeptical about. You may be told sequencing has “five times the yield” of a microarray. That kind of comparison is put together by placing figures from separate studies of different populations side by side, and it circulates mainly in laboratory marketing. In the one head-to-head study on this page, the authors concluded that the yields of the two tests were comparable — 9.3% for microarray across all 258 children, 8.4% for exome in the 95 who had it (Tammimies 2015). The honest case for sequencing is not that it is five times better; it is that it reads a different and larger part of the genome, that guidelines now support it early, and that combining approaches finds more than either alone.

What the result can say

There are three broad answers, and it helps to know all three before the phone rings.

A positive result. The lab has found a change it can classify as pathogenic or likely pathogenic, and considers it sufficient to explain part of your child’s picture. You should get a named condition, a written report and — this is the part to insist on — a plan.

A negative result. Nothing reportable was found. This is the most common outcome. It does not mean there is no genetic contribution, only that this test, reading this part of the genome, with today’s knowledge, found nothing it could interpret. Ask whether re-analysis of the stored data is planned, and when. Genes are described every year that were unknown when your child’s sample was run.

An uncertain result. The ACMG and the Association for Molecular Pathology set out a five-tier classification for sequence variants — pathogenic, likely pathogenic, uncertain significance, likely benign, benign — with criteria for placing a variant in each (Richards 2015). A variant of uncertain significance, usually shortened to VUS, sits in the middle. The lab found a change and cannot yet say whether it matters.

Uncertain results are common, and they are the ones families are least prepared for. A VUS is not a diagnosis, not a warning, and not something to act on medically by itself. What it is, is a reason to ask three questions: what would move this variant into a firm category, is anyone looking, and what should we do in the meantime? Usually the answer to the last is: nothing different. There is more on reading a report in what your results mean.

Trio testing

Testing your child alone is called proband-only. Testing your child alongside both parents is called a trio. A trio lets the lab see which changes your child inherited and which are new, which is what resolves many otherwise uncertain results. It is worth asking whether trio testing is available and covered before the sample is taken, because adding parents afterwards means starting parts of the process again.

Secondary findings

Because sequencing reads thousands of genes at once, it can turn up something unrelated to the reason you tested. The ACMG maintains a list of genes considered actionable enough to report when found this way; the SF v3.3 policy statement, updated in 2025, names 84 (Lee 2025). Most are tied to heart conditions — rhythm disorders, cardiomyopathies, aortic disease and inherited high cholesterol — or to inherited cancer predisposition, places where knowing early genuinely changes what doctors watch for.

Receiving them is a choice, not an automatic extra: labs generally ask before testing begins, so ask what your lab’s process is. There is no correct answer. Some parents want every actionable finding; others want only what explains the reason they came. Make it deliberately rather than by ticking whatever is already ticked, and if your child is old enough to have a view, ask them.

What a result changes

This is the section to read first if you are deciding whether testing is worth the appointment.

Screening schedules. This is the biggest practical gain. Many named genetic conditions come with recommended surveillance — echocardiograms, kidney ultrasounds, eye checks, hearing checks, or watching for seizures — because those problems occur often enough in that condition to look for them before they announce themselves. The 2025 AAP clinical report names surveillance for disorder-related complications among the benefits of establishing a specific diagnosis early (Rodan 2025). Without a name, nobody knows to look.

Condition-specific management. Occasionally a result points to something more specific than surveillance: a medication that suits or does not suit that condition, a dietary consideration, an anesthetic precaution worth putting on the chart before a routine procedure. This is uncommon, and it is real when it happens.

Research and registries. A confirmed genetic finding is the key to studies that are otherwise closed. SPARK is a large US autism research community for people of any age living in the US with a professional diagnosis of autism, and their family members. Its FAQ says “There is no cost to join SPARK”, that enrollment “usually takes between 15 and 30 minutes”, that the saliva kit for DNA analysis is optional, and that “If we find a genetic change related to autism in your sample, we will contact you.” Simons Searchlight is narrower: it follows families with a confirmed diagnosis of one of the rare genetic neurodevelopmental conditions on its gene list, asks participants to upload their genetic lab report, and covers the cost of any sample collection it asks for. Neither is treatment, and neither is a substitute for care. What they offer is a way for your child’s information to count toward what is known about their condition, and often a route to other families with the same one.

Information for your family. Some findings carry a recurrence risk — a known chance of the same thing happening in another pregnancy — and some do not. The 2025 AAP report lists accurate recurrence risk as one of the benefits of establishing a specific diagnosis (Rodan 2025). This is genetic counselor territory, not internet territory, and there is no rush. The information keeps.

School: usually very little. Be prepared for this one. Educational support is driven by what your child needs day to day, not by a gene name. A genetic result rarely changes classroom strategy. It can occasionally help a school understand medical absences, fatigue or a seizure plan, and it can help an argument about whether needs are real — but if the reason you are testing is to get better support at school, testing is an indirect route to a goal that an evaluation and the right diagnostic codes address more directly.

Your child’s own story, later. A named condition is something your child may want, or may not, when they are old enough to ask why autism happens in their own case. Some autistic adults value having the explanation. Others find it beside the point. Both reactions are normal, and neither is yours to decide for them now — which is a good reason to keep the report somewhere they can find it, and to be honest with them about what it does and does not say.

What it costs and who pays

Cost is the question families ask last and worry about first. The honest answer is that the range is enormous and the sticker price is rarely what you pay.

MedlinePlus, the National Library of Medicine’s consumer genetics resource, puts it plainly: the cost of genetic testing “can range from under $100 to more than $2,000, depending on the nature and complexity of the test.” A single targeted test sits at the bottom of that range and sequencing at the top, and testing more than one family member — a trio, for instance — raises the total.

That is list price. What an insured family actually pays depends on the plan, on whether prior authorization was obtained, and on whether the lab has a patient-assistance program. It is common for a bill that looks alarming on paper to settle at a fraction of it. It is also possible for a test run without authorization to land in full on a family that assumed it was covered, which is the outcome to protect yourself against.

Three things make coverage more likely. First, a letter of medical necessity from the ordering clinician, setting out the clinical reason for the specific test. Second, prior authorization obtained in writing before the sample goes to the lab. Third, matching the request to the wording insurers recognize — sequencing is generally easier to get approved where there is developmental delay or intellectual disability, which is exactly the population the 2021 ACMG guideline addresses (Manickam 2021).

If your child has Medicaid, check your own state’s policy rather than a national summary: Medicaid genetic-testing coverage is set state by state and varies considerably. Our fuller walkthrough of billing, denials and appeals is on what genetic testing costs and how insurance handles it.

How to get tested

Who can order it. A clinical geneticist or a genetic counselor, a child neurologist, or a developmental pediatrician. A general pediatrician can often order a microarray or fragile X test directly and can always write the referral to genetics. If your child has already seen a neurologist for seizures, that clinic is frequently the fastest route.

What to say. You do not need technical language, and you do not need to justify the request. Something like this is enough:

“My child has an autism diagnosis. I would like a referral to genetics to discuss testing. I understand that professional guidelines support exome or genome sequencing early where there is developmental delay, and I would like to talk through what would be appropriate for us — including whether fragile X needs its own test.”

If you would rather not say it out loud, put it in the patient portal in writing. A written request also gives you a dated record if the referral goes missing.

What to bring. The diagnostic report, any previous genetic test results — including ones you think were negative, since re-running a test is a waste of everyone’s time — a list of medications, and a short family history: developmental conditions, intellectual disability, epilepsy, miscarriages, early deaths, or relatives who were “always a bit different”. Bring names and rough ages. Also bring your questions, on paper, because appointments move fast.

Saliva or blood. Both are used. Blood usually yields more DNA, and some labs require it for sequencing; saliva or a cheek swab avoids a needle entirely and can sometimes be collected at home. If a needle would make the appointment impossible, say so before it is booked and ask whether a saliva kit is acceptable for the specific test being ordered — for many, it is. If blood is genuinely required, the preparation list further down this page is the one to use. There is more on what is and is not drawn from a sample in blood tests and autism.

Turnaround. MedlinePlus says that from the date a sample is taken, “it may take a few days to weeks to receive the test results”, and that the clinician or genetic counselor who orders a particular test can tell you the time frame for that one. Ask two things at the appointment: what turnaround the lab quotes, and who will call you. A result sitting unexplained in a patient portal on a Friday night is avoidable, and asking for it to come by phone is a reasonable request.

Direct-to-consumer kits are not this. The ancestry and wellness kits sold online are not clinical genetic testing. They read a limited, pre-chosen set of positions in the genome, they are not designed or validated to diagnose conditions, and their results are not a substitute for a clinical report — a raw data file from one of them cannot answer any of the questions on this page. If you have already bought one, nothing is lost; just do not let it stand in for the real test.

The form at the foot of this page registers your interest with a genetics or testing partner — not with us — if you would like help finding a route to testing rather than doing all the chasing yourself. It is not a referral, no specific lab is chosen for you, and nothing on this page depends on it.

Deciding not to test

Some families read everything here and decide not to test. That is a legitimate decision, and it is not a failure of diligence.

The reasons are usually good ones. Most results find nothing, so the most likely outcome is an appointment, a needle and a wait for no new information. Uncertain results can create worry that outlasts their usefulness. Some parents do not want a genetic file attached to their child’s name. Some autistic adults are wary of the whole field, for reasons rooted in how genetic knowledge about autism has been used, and a parent who shares that wariness is not being irrational.

Autistic people have the right to decide about their own genetic information. For a young child that decision is held by parents, but it is being held in trust, not owned — and it should get lighter to hold every year. A teenager who does not want to be tested has said something that matters. A teenager who wants to be tested when you would rather not has also said something that matters.

None of this is permanent. Testing that is declined now can be done later; the biology does not expire. If you want a middle path, ask genetics to see you for the conversation without committing to the test — the 2013 ACMG guideline explicitly expects testing options, diagnostic yields and family investment to be discussed before anyone proceeds (Schaefer 2013). Having the conversation and then saying no is a complete and reasonable outcome.

Questions to ask genetics

1. What exactly are you ordering, and what does that test read?
2. How often do you find something in a child like mine — and which study is that number from?
3. Has fragile X been ordered separately?
4. Is trio testing available, and is it covered?
5. What would change in my child’s care in the next year if this comes back positive?
6. How do you handle secondary findings, and when do we make that choice?
7. If the result is uncertain, who explains it to us, and is re-analysis planned later?
8. What will this cost us, and has prior authorization been obtained in writing?

Write the answers down. Future clinicians will ask you the same things, and you will not remember which appointment they came from.

Preparing for the blood or saliva draw

• Ask for the first or the last appointment of the day, when waiting rooms are quietest.
• Ask whether saliva is acceptable for this specific test before assuming a needle is required.
• Request numbing cream and ask how long before the appointment it needs to go on.
• Pack headphones, a preferred video and a snack for afterwards.
• Tell the staff in advance how your child communicates, and what a distressed moment looks like for them.
• Ask for the draw to be done by the most experienced person available, and ask them to explain each step before doing it.
• Plan nothing else for that day. One appointment is enough.

Our fuller accommodation list for clinical appointments is on medical appointments.

Questions parents ask after reading this

1. Given who my child actually is — other medical features or none — which yield figure on this page describes us, and does that change whether we test?
2. If the result came back uncertain, what would we do differently the next morning? If the answer is nothing, are we ready for that outcome?
3. Who in this family gets to decide, and how does that shift as our child gets older?
4. Are we testing to change medical care, to answer a question we have, or to get something at school — and is testing the right tool for the one we picked?

If you only do three things

1. Ask your pediatrician, in writing, whether a genetics evaluation has been offered — and ask specifically whether fragile X has been ordered separately.
2. Before consenting, ask what would change in your child’s care in the next year if the result were positive, and write the answer down.
3. Decide in advance, with your co-parent and with your child if they are old enough, what you want to do about secondary findings — so you are not deciding it at a desk with a pen in your hand.

Related

Whole exome sequencing · Chromosomal microarray · Fragile X · What testing costs · Understanding your results · Autism testing, explained · Getting assessed · Why autism happens