Autism rates have risen dramatically over the past two decades, but the increase reflects several overlapping factors rather than a single cause. In 2000, the CDC estimated that 1 in 150 children had autism. By 2022, that number had reached 1 in 31. Some of that rise comes from genuine biological risk factors, but a large portion stems from changes in how autism is defined, who gets screened, and which children clinicians now recognize.
The Definition of Autism Got Broader
Before 2013, what we now call autism was split into five separate diagnoses. A child could receive a label of autistic disorder, Asperger syndrome, or “pervasive developmental disorder not otherwise specified” (PDD-NOS), among others. Each had its own criteria, and many children on the milder end never received any diagnosis at all.
When the DSM-5 was published in 2013, all of those categories were collapsed into a single label: autism spectrum disorder. This meant that children who previously would have been diagnosed with Asperger syndrome or PDD-NOS were now counted in the same category as children with more significant support needs. That consolidation alone pushed the official numbers upward without a single additional child actually developing the condition.
Diagnostic Substitution Plays a Role
As autism diagnoses rose through the 1990s and 2000s, diagnoses of intellectual disability and learning disabilities fell. Between 1994 and 2003, the prevalence of intellectual disability dropped by 2.8 per 1,000 students, and learning disabilities dropped by 8.3 per 1,000. Those declines tracked closely with the rise in autism, suggesting that many children who would have previously been categorized under a different label were now being identified as autistic instead. This phenomenon, called diagnostic substitution, means part of the increase is a reshuffling of categories rather than new cases appearing.
The special education system tells a similar story. In the 2008-09 school year, 4.97% of students with disabilities were identified under the autism category. By 2023-24, that had risen to 13.67%. Today, roughly 876,000 students receive services under the autism classification. Some of these students always had autism; the system simply lacked the right label for them.
Screening Catches Children Earlier
Pediatricians now routinely screen toddlers for autism using standardized tools between 16 and 26 months of age. A generation ago, this kind of universal screening didn’t exist. Children who were verbal and academically capable could go unnoticed for years, or never be flagged at all.
Research from Children’s Hospital of Philadelphia found that children who screened positive on early screening tools were diagnosed seven months earlier than those who screened negative but were later identified. Earlier diagnosis doesn’t create more autism, but it does increase the number of children who appear in prevalence statistics at younger ages, making the overall count rise faster.
Girls Are Finally Being Recognized
Autism was historically framed as a male condition. Early researchers described it as a “variant of male intelligence,” and diagnostic criteria were built around how boys typically present. Girls who masked their social difficulties or whose traits looked different from the male prototype were frequently missed.
That gap is narrowing. Over a 20-year period tracked in North Carolina, the ratio of males to females receiving an autism diagnosis dropped from 5.64 to 1 down to 3.07 to 1. The CDC’s national estimate has similarly shifted from 4.2 to 1 in 2016 to 3.8 to 1 more recently. Clinicians are now better trained to spot autism in girls, and the DSM-5 removed a previous requirement for language delay that had disproportionately excluded females. Every girl who gets identified today but would have been missed in 2004 adds to the prevalence count.
Genetics Account for Most Individual Risk
Twin studies consistently show that genetics are the dominant factor in whether a specific child develops autism. A large meta-analysis of twin research estimated heritability between 64% and 91%, with a central estimate around 74%. In practical terms, if one identical twin is autistic, the other is far more likely to be autistic than would be the case for fraternal twins, pointing strongly to inherited biology.
Hundreds of genes contribute to autism risk, and most cases involve complex combinations rather than a single gene mutation. This genetic architecture hasn’t changed over a few decades, which is one reason researchers believe the rise in prevalence is largely driven by identification rather than a surge in new biological cases.
Parental Age and Prenatal Exposures
Some biological risk factors have genuinely shifted at the population level. The average age at which people have children has climbed steadily, and paternal age is one of the more consistent predictors of autism risk. A study published in JAMA Psychiatry found that children born to fathers aged 40 or older were 5.75 times more likely to be diagnosed with autism compared to children of fathers under 30, even after adjusting for maternal age and socioeconomic factors.
Prenatal exposure to air pollution has also drawn attention. Research published in JAMA Network Open identified specific components of fine particulate matter, particularly sulfate and ammonium, as associated with higher autism risk during pregnancy. Children exposed to higher levels of ammonium during prenatal development had a 12-13% increase in risk per interquartile-range increase in exposure. These are modest effect sizes at the individual level, but across millions of pregnancies, even small shifts in environmental exposures could nudge population-level rates.
Preterm birth is another factor. Babies born prematurely have higher rates of autism compared to those born at term, and advances in neonatal medicine have dramatically improved survival for extremely premature infants. Children who a generation ago would not have survived are now thriving, and some carry neurodevelopmental differences including autism.
How Much of the Increase Is “Real”?
This is the question researchers still debate. The honest answer is that both things are true simultaneously: we are identifying autism in people who always had it but were overlooked, and some genuine risk factors have become more common. Broader diagnostic criteria, universal screening, better recognition of girls, and diagnostic substitution from other categories account for a substantial portion of the rise. Meanwhile, older parental age, improved preterm survival, and environmental exposures likely contribute a smaller but real biological component.
No serious researcher attributes the increase to vaccines. That hypothesis was based on a single fraudulent study that has been retracted, and dozens of large-scale studies involving millions of children have found no connection. The actual story is more complicated and less alarming: we got better at recognizing a condition that was always more common than we realized, while a few modifiable risk factors shifted at the margins.