Autism does not develop after birth. It is a neurodevelopmental condition whose biological foundations are laid during pregnancy, when the brain is first forming. What can happen after birth is that autism becomes *visible*, as a child grows and faces social and communication demands they struggle to meet. This distinction between when autism originates and when it’s recognized is the source of much confusion.
Autism Begins in the Womb
The fetal period is when neurons are produced, migrate to their correct locations, and differentiate into specialized brain regions. Disruptions to this process, whether from genetic mutations or environmental exposures, can alter the trajectory of brain development in ways consistent with autism. Animal studies illustrate this clearly: when researchers exposed developing brains to valproate (a compound known to increase autism risk) during pregnancy, they found distinct cellular changes in key brain areas involved in social behavior, decision-making, and motor coordination. When the same exposure happened after delivery, those changes did not occur.
Genetics plays a major role. About half of autism liability comes from common genetic variation, the kind inherited from parents. Another 16 to 19 percent comes from rare mutations, many of which arise spontaneously in the child rather than being passed down. These spontaneous mutations, called de novo mutations, affect genes active in early brain development. Carrying certain types can increase individual risk by roughly five times compared to other inherited variants.
Prenatal Risk Factors
A range of environmental exposures during pregnancy are associated with increased autism risk, and the specifics matter. Viral infections in the first trimester that require hospitalization, including influenza and gastroenteritis, are linked to a 2.8-fold increase in risk. Bacterial infections in the second trimester carry a smaller but still notable 1.4-fold increase. Fever during pregnancy on its own, independent of the type of infection, doubles the odds, though mothers who took fever-reducing medication showed lower risk, suggesting the fever itself may be part of the mechanism.
Maternal obesity before pregnancy increases the odds by 1.3 to 2.0 times. Gestational diabetes raises risk by 1.5 to 1.7 times, with earlier diagnosis (before 26 weeks) carrying a higher association. Excessive weight gain during pregnancy adds a smaller but measurable increase of 1.1 to 1.6 times. Use of certain antidepressants during pregnancy, particularly in the first trimester, has been linked to a 1.5 to 4.5-fold increase, though separating the effect of the medication from the underlying condition it treats remains an active question.
None of these factors guarantee autism. They shift probability during a window when the brain is uniquely vulnerable to disruption.
Why It Seems to Appear After Birth
Many parents notice the first signs of autism between ages 1 and 3, which can create the impression that something happened to cause it. In reality, the brain differences were already present. They simply become apparent as children reach developmental milestones that require social communication, flexible play, and language, skills that depend on the very brain systems affected.
Researchers have identified subtle behavioral markers in infants well before a typical diagnosis. By 6 months, some high-risk infants show atypical neural responses to social cues like eye contact. Between 2 and 6 months, reduced fixation on people’s eyes has been documented in infants later diagnosed with autism. By 7 months, differences in social referencing (looking to a caregiver for cues about how to react) and visual tracking can distinguish infants who will later receive a diagnosis from those who won’t. These are not signs a parent would easily spot, but they confirm that autism is already shaping behavior long before it becomes obvious.
Despite these early markers, the median age of diagnosis in the United States is 47 months, nearly 4 years old, and ranges as high as nearly 6 years in some regions. Autism can be reliably identified in some children as early as 12 months, but most cases are caught much later.
Regression and Skill Loss
About one-third of young children with autism lose skills they previously had, most often speech, but sometimes social engagement or play abilities as well. This regression typically happens in the second or third year of life, with an average onset around 20 months. For parents, watching a child who was babbling or making eye contact suddenly stop can feel like proof that something went wrong after birth. But regression in autism is understood as a manifestation of an already-different neurodevelopmental trajectory, not a new injury to a previously typical brain.
A small number of parents report skill loss after age 2. In rare cases, a more severe and global regression occurs after several years of entirely normal development. This pattern was historically called childhood disintegrative disorder. It begins later, involves more dramatic loss across multiple domains, and typically carries a worse prognosis. It is now classified under the broader autism spectrum, but its distinct presentation and timeline set it apart from the more common pattern of early regression.
Late Diagnosis Is Not Late Onset
Some people are not diagnosed with autism until adolescence or adulthood, which can raise the question of whether their autism somehow developed later. It didn’t. A subgroup of autistic people, particularly those with strong intellectual abilities, consciously develop compensatory strategies that mask their difficulties. They may maintain good eye contact, mimic social reciprocity, and suppress visible restricted interests, appearing neurotypical in many contexts.
This compensation disguises but does not eliminate the underlying differences. The result is that some individuals go decades without a diagnosis, missing out on support and accommodations. This pattern is especially common in autistic women, who tend to compensate more extensively than men. A person diagnosed at 30 was autistic at 3. The diagnosis is what’s new, not the condition.
What Doesn’t Cause Autism
Because autism becomes apparent in the same developmental window when children receive routine vaccinations, a persistent myth links the two. Large-scale studies across multiple countries have repeatedly found no connection between vaccines and autism. Parenting style also does not cause autism. The “refrigerator mother” theory, which blamed cold or unresponsive parenting, was thoroughly debunked decades ago. Positive parenting can support an autistic child’s development, but its absence does not create the condition.
The scientific consensus is clear: autism’s origins are prenatal, rooted in a combination of genetic susceptibility and, in many cases, environmental exposures during pregnancy. Postnatal life shapes how autism is expressed, how well a person is supported, and when it’s identified. But it does not create autism where the neurological foundation didn’t already exist.