What Percentage of the US Population Has Autism?

About one in 31 eight-year-old children in the United States had autism spectrum disorder (ASD) in 2022, according to the most recent surveillance data from the CDC. That translates to roughly 3.2% of children at the age most commonly tracked. For adults, the picture is murkier, but one modeling study estimated that about 2.2% of U.S. adults were living with autism in 2017. These figures have risen sharply over the past two decades, and the reasons behind that rise involve diagnosis, awareness, and access to services as much as any change in biology.

The Latest Childhood Numbers

The benchmark figure in the U.S. comes from the CDC’s Autism and Developmental Disabilities Monitoring (ADDM) Network, which tracks autism prevalence among eight-year-olds across multiple sites. The most recent report, covering 2022, found an overall prevalence of 32.2 per 1,000 children, or one in 31, across 16 surveillance sites. Site-level estimates ranged dramatically, from about 10 per 1,000 in one Texas site to 53 per 1,000 in California.¹ That fivefold spread between sites within the same country tells you immediately that what gets counted depends heavily on where and how you count.

To put the trajectory in perspective: in 2010, the ADDM Network reported a prevalence of about one in 68 among eight-year-olds.² By 2014, it was one in 59.³ A decade later, one in 31. The number has roughly doubled in twelve years. Whether that represents a genuine increase in how many children have autism or an expansion in who gets identified is one of the most debated questions in developmental medicine.

How Many Adults Have Autism

Counting autistic adults is harder than counting children, because most of the surveillance infrastructure was built around kids. There is no adult equivalent of the ADDM Network. One widely cited modeling study estimated that about 2.2% of U.S. adults aged 18 to 84, roughly 5.4 million people, were living with ASD in 2017. State-level estimates in that analysis ranged from about 2.0% in Louisiana to 2.4% in Massachusetts.⁴ That study used simulation modeling rather than direct clinical assessment, so the numbers carry real uncertainty. But the finding suggests millions of American adults are autistic, many of them undiagnosed.

Diagnosis data paints a different picture from modeling. An analysis of U.S. claims and survey data from 2011 to 2022 found that documented autism diagnoses drop off steeply with age. Among five-to-eight-year-olds in 2022, about 30 per 1,000 carried a diagnosis. Among people aged 45 to 54, it was just 0.6 per 1,000, and for those 65 and older, 0.2 per 1,000.⁵ Those tiny numbers for older adults don’t mean autism is rare in that generation. They reflect decades when autism was understood almost exclusively as a severe childhood condition, when Asperger syndrome wasn’t yet part of the diagnostic lexicon, and when many people simply never got evaluated. A separate look at adults enrolled in Medicaid found the diagnosed prevalence among all ages more than doubled from about 4.2 per 1,000 enrollees in 2011 to 9.5 per 1,000 in 2019.⁶ That rapid climb in adult diagnoses is happening in real time.

Why the Numbers Keep Climbing

Every time the CDC releases a new prevalence estimate, headlines announce that autism is “surging.” The reality is more complicated. A significant share of the increase appears to come from better identification of people who would always have met criteria but were previously overlooked or labeled with something else.

The strongest piece of evidence for this is the changing profile of who gets counted. Between 2000 and 2016, the prevalence of autism among children with mild, borderline, or no adaptive challenges rose from about 5 per 1,000 to nearly 18 per 1,000. During the same period, the prevalence of autism with moderate-to-profound adaptive challenges actually declined slightly, from about 1.5 to 1.2 per 1,000.⁷ In other words, virtually all of the growth came from identifying children with milder forms of autism. The number of children with more pronounced support needs barely changed.

Diagnostic substitution is another contributor. A California-based study found that roughly a third of the increase in autism prevalence between 1996 and 2004 in special education records could be attributed to children switching from another disability classification to autism.⁸ However, this explanation has limits. A broader state-by-state analysis found that intellectual disability rates stayed relatively flat in most states, including the most populous ones, even as autism prevalence rose sharply. The hypothesis that children were simply being relabeled from intellectual disability to autism didn’t hold up well once researchers looked beyond aggregated national data.⁹

Changes to the diagnostic manual also matter. When the DSM-5 replaced the DSM-IV-TR in 2013, it folded several previously separate diagnoses (autistic disorder, Asperger’s disorder, pervasive developmental disorder not otherwise specified) into a single autism spectrum disorder category. Studies comparing the two systems have found that roughly 81% of children classified under the older criteria also met DSM-5 criteria, but the rest did not, and the DSM-5 appeared to yield a slightly lower overall prevalence for the same population.¹⁰ Other reviews found a wider range, with anywhere from 9% to 54% of people diagnosed under DSM-IV criteria not meeting DSM-5 thresholds.¹¹ In practice, though, the DSM-5 transition coincided with continued prevalence increases, suggesting that broader awareness and screening more than compensated for any tightening of criteria.

The Sex Gap Is Narrower Than People Think

Autism has long been described as a condition affecting boys far more than girls, with a commonly cited ratio of four males for every female. That ratio overstates the gap. A large meta-analysis pooling over 13 million participants found the overall male-to-female ratio was about 4.2 to 1, but in higher-quality studies the ratio dropped to about 3.3 to 1. When researchers actively screened the general population instead of relying on existing diagnoses, the ratio fell further to about 3.2 to 1.¹² The conclusion: girls who meet the criteria for autism are disproportionately likely to go undiagnosed.

Newer data from Sweden pushes this further. A large birth-cohort study tracking diagnoses through 2022 found that the male-to-female ratio decreased with increasing age at diagnosis. By age 20, the cumulative ratio had fallen to about 1.2 to 1. The researchers projected that at the rate the gap was closing, it would reach parity by 2024.¹³ Whether these Swedish findings will hold in American populations remains to be seen, but they reinforce the view that the traditional male-skewed picture partly reflects who gets noticed and referred rather than who has autism. Diagnosis data from the U.S. still shows about a three-to-one male-to-female ratio overall.⁵

Racial and Ethnic Patterns Have Flipped

For years, autism prevalence in the U.S. was highest among White children. That pattern has reversed. In nearly all birth cohorts before 2010, White children had the highest rates. Starting with the 2010 birth cohort, prevalence among Black, Hispanic, and Asian/Pacific Islander children surpassed rates among White children and continued climbing.¹⁴ The 2020 ADDM data showed ASD prevalence among White eight-year-olds at about 24.3 per 1,000, compared with 29.3 for Black children, 31.6 for Hispanic children, and 33.4 for Asian/Pacific Islander children.¹⁵

This reversal isn’t because autism became more common among minority children in some biological sense. Earlier disparities likely reflected unequal access to diagnosis. White families, especially in higher-income areas, had easier paths to evaluation, specialist referrals, and school-based services. As screening became more systematic and awareness campaigns reached more communities, children of color started getting identified at rates that probably better reflect the true distribution. A study tracking this shift directly found that by 2021, the prevalence of diagnosed ASD among Black children (about 4.0%) had overtaken that of White children (about 3.9%), after trailing them as recently as 2017.¹⁶

Income and Geography Shape Who Gets Counted

Where you live and how much your family earns still influence the likelihood of an autism diagnosis. In the U.S., prevalence has consistently been higher in higher-income areas. One study found that autism prevalence in census tracts with median incomes above $90,000 was about 17 per 1,000, compared with roughly 7 per 1,000 in tracts with median incomes at or below $30,000. The prevalence ratio between the highest and lowest income areas was about 2.2 to 1, even after adjusting for race and ethnicity.¹⁷ This socioeconomic gap persisted across surveillance years and across racial groups.¹⁸

Geographic variation adds another layer. State-level prevalence of ever-diagnosed ASD among children ranged from about 1.5% in Texas to nearly 4.9% in Florida, based on 2016 data.¹⁹ A finer-grained analysis found regional differences in prevalence ranging from about 4.4% in the Mid-Atlantic to 2.7% in the West South-Central region.²⁰ These gaps likely reflect differences in the density of developmental pediatricians, the availability of school-based screening, state Medicaid policies, and cultural attitudes toward seeking a diagnosis. They are less about where autism occurs and more about where it gets noticed.

How the U.S. Compares Globally

The U.S. reports higher autism prevalence than most other regions. A three-level meta-analysis of studies worldwide found a pooled prevalence of about 1.0% in North America, compared with roughly 0.7% in Europe and 0.4% in Asia.²¹ These differences almost certainly reflect variation in diagnostic infrastructure and cultural thresholds for referral more than genuine biological differences across continents. Countries with well-resourced developmental screening systems tend to report higher prevalence, which is exactly what you’d expect if a large share of autism goes undiagnosed where services are scarce.

Co-occurring Intellectual Disability

Not everyone with autism has intellectual disability, but the overlap is substantial and has implications for service needs. Among children with ASD and available cognitive data in the 2020 ADDM Network, about 38% were classified as having intellectual disability. That proportion varied by race: roughly half of Black children with ASD had co-occurring intellectual disability, compared with about 32% of White children. Children with intellectual disability were also diagnosed with autism earlier, at a median age of about 43 months, compared with 53 months for those without intellectual disability.¹⁵ These numbers matter because they affect what kinds of support a person needs across their lifetime, from school accommodations to adult housing.

When Autism Gets Diagnosed

Reliable signs of autism can be present by 18 months, and the American Academy of Pediatrics recommends screening at the 18- and 24-month well-child visits. Despite this, the average age at diagnosis remains stubbornly high. A systematic review of studies from 2012 to 2019 found a mean age at diagnosis of about 60 months overall, and about 43 months when the analysis was restricted to studies of children under 10.²² That means many children are entering kindergarten before they receive a formal diagnosis, which delays access to early intervention services that have the strongest evidence base for improving long-term outcomes.

The gap between when autism could be detected and when it actually is detected widens for girls, for children in lower-income families, and for children of color. These are the same groups where overall prevalence numbers appear to be catching up, suggesting that what looks like “rising rates” is partly diagnosis reaching populations it was previously missing.

The Genetics Behind Prevalence

One reason researchers are cautious about attributing the entire prevalence increase to environmental change is that autism is highly heritable. A large multinational study covering over two million children across five Nordic countries estimated heritability at about 80 to 83%, depending on the statistical model used. Shared environmental effects were minimal, with estimates hovering near zero in most models. Nonshared environmental effects accounted for roughly 17 to 18%.²³ That high heritability figure doesn’t mean a single “autism gene” exists. Hundreds of genetic variants contribute, each with a small effect, and most cases arise from common genetic variation spread across many genes rather than from rare mutations.

High heritability doesn’t rule out environmental contributions. It means that within current populations and current environmental variation, genetic differences account for most of the variation in who develops autism. If the environment changed dramatically for everyone, those proportions could shift. Parental age at conception, prenatal exposures, and complications during birth have all been associated with increased autism risk in epidemiological studies, but their individual contributions are small compared with genetic factors.

The Economic Weight of These Numbers

As prevalence figures climb, the financial implications compound. A study comparing families of children with and without autism found that ASD was associated with roughly $3,000 in additional annual healthcare costs and about $14,000 in additional non-healthcare costs, with school systems absorbing the largest share at around $8,600 per year.²⁴ Scaled nationally, the combined annual cost of autism-related direct medical, non-medical, and productivity losses was forecast at about $268 billion for 2015, and projected to reach $461 billion by 2025 if trends continued.²⁵ Those figures approach the cost burden of diabetes and exceed estimates for stroke and hypertension.

Lifetime costs are even more striking. One estimate placed the average lifetime social cost per person with autism at roughly $3.6 million. Accumulated across the U.S. population through 2029, total lifetime costs could reach between $11.5 and $15 trillion, depending on whether prevalence continues to rise.²⁶ These numbers reflect not just healthcare but lost productivity, special education, residential care, and family caregiving. They underline why the prevalence question matters beyond epidemiology — every fraction of a percentage point translates into billions of dollars in service demand.

Mortality and Life Expectancy

The prevalence question implicitly raises another one: what happens to autistic people over a lifetime? The data here is limited but sobering. A 20-year follow-up of autistic adults found that those who died did so at an average age of 39, dying an average of nearly 39 years before their expected life span. Most deaths occurred during the person’s 20s, 30s, or 40s.²⁷ This was a small sample and likely skewed toward individuals with higher support needs, but it points to a real disparity. A larger analysis of hospital discharge data found that autistic adults had about 50% higher odds of dying during a hospitalization compared with matched controls, even after adjusting for age, sex, income, and co-occurring conditions.²⁸ Epilepsy, accidents, and difficulty communicating symptoms in medical settings all contribute to elevated mortality risk. As the population of diagnosed autistic adults grows, understanding and addressing these risks becomes more urgent.