Autism starts long before anyone notices it. The biological roots trace back to prenatal brain development, with most risk genes active during the first and second trimesters, but the behavioral signs that parents and clinicians recognize tend to surface between 12 and 24 months of age. Formal diagnosis lags even further behind, arriving for most children somewhere between ages 3 and 5. That gap between biological onset, visible signs, and official diagnosis shapes nearly everything about how autism is understood, detected, and treated.
Autism Begins in the Womb
The idea that autism “starts” at any particular age after birth is, in one sense, misleading. Research over the past decade has made a strong case that the foundational differences in brain wiring begin during pregnancy. Most genes linked to autism risk are expressed prenatally, and they tend to fall into two waves. During the first two trimesters, one set of genes disrupts early processes like cell proliferation, the generation of new neurons, and how those neurons migrate to their correct positions. During the third trimester and into early postnatal life, a second set disrupts the formation of connections between neurons and the wiring of the cortex.1PubMed Central. Prenatal Origins of ASD: The When, What, and How of ASD Development These are not two separate problems; they are stages of the same developmental process going differently.
Physical evidence supports this timeline. One study found that head circumference in fetuses later diagnosed with autism was already measurably larger during the second and third trimesters, beginning around the 22nd week of pregnancy, a period that matters for cortical layering and glial cell activity.2PubMed. Autism is a prenatal disorder: Evidence from late gestation brain overgrowth Another study linked fetal brain measurements at 20 and 28 weeks of gestation to scores on an autism traits questionnaire completed after birth, finding that larger cerebellar measurements at 20 weeks and larger head circumference at 28 weeks were each associated with higher trait scores.3PubMed Central. Fetal brain growth and infant autistic traits
Environmental exposures during pregnancy add another layer. Maternal infection, gestational diabetes, and obesity are established risk factors.4PubMed Central. Prenatal environmental risk factors for autism spectrum disorder and their potential mechanisms Air pollution exposure during specific gestational windows has also drawn attention. A large study identified weeks 14 through 32 of pregnancy as a sensitive window for fine particulate matter exposure, and the third trimester as particularly relevant for certain pollutant components.5JAMA Network Open. Prenatal Exposure to Fine Particulate Matter Components and Autism Risk in Childhood Separate research in the eastern United States found results consistent with increased susceptibility during the third trimester as well.6Epidemiology. Particulate Matter Exposure, Prenatal and Postnatal Windows of Susceptibility, and Autism Spectrum Disorders None of this means a single exposure causes autism; rather, these environmental factors interact with genetic vulnerability during windows when the developing brain is especially sensitive.
What Happens in the First Year
If the prenatal groundwork is already laid, what changes after birth? The differences become measurable through brain imaging and, to some extent, through behavior, but they are subtle enough that most parents and pediatricians would not pick up on them during routine checkups.
Brain growth tells part of the story. A landmark neuroimaging study followed over a hundred high-risk infants and found that those later diagnosed with autism showed an unusual expansion of cortical surface area between 6 and 12 months, followed by brain volume overgrowth between 12 and 24 months. The volume overgrowth was linked to the emergence and severity of social difficulties.7PubMed Central. Early brain development in infants at high risk for autism spectrum disorder Earlier work had shown a similar pattern using head circumference: infants later diagnosed with autism actually had slightly smaller heads at birth, but their head growth accelerated dramatically in the first six to fourteen months. Roughly 59% of those with autistic disorder showed accelerated growth trajectories during that period, compared to only about 6% of typically developing infants.8JAMA. Evidence of Brain Overgrowth in the First Year of Life in Autism
Brain activity patterns also diverge early. One study using EEG was able to distinguish infants later diagnosed with autism from low-risk controls with near-perfect accuracy starting at three months of age.9Scientific Reports. EEG Analytics for Early Detection of Autism Spectrum Disorder: A data-driven approach Other EEG research found that the normal sequence of brain rhythm maturation was preserved in autism but happened at a faster rate, with a steeper trajectory between 3 and 12 months.10PubMed Central. Accelerated Infant Brain Rhythm Maturation in Autism The brain is developing along a recognizable path, but the timing is off.
Behavioral differences during the first year are harder to pin down reliably. Eye-tracking research has shown that reduced looking time at people and faces, along with problems disengaging attention, are among the earliest observable signs, sometimes visible within the first year.11PubMed Central. Eye tracking in early autism research However, researchers have also found that some proposed markers do not hold up as cleanly as hoped. One study testing whether infants’ responses to direct eye gaze could predict later autism found that while infants with a later diagnosis tended to look away from faces with direct gaze more quickly than typical-likelihood infants, this measure did not reliably differentiate those who would go on to be diagnosed from those who would not.12PubMed Central. Infant responses to direct gaze and associations to autism: A live eye-tracking study The evidence for first-year behavioral biomarkers is promising but uneven.
Even infant cries have been studied as a potential early signal. Research suggests that atypical vocal calls in infancy, including differences in fundamental frequency, cry duration, and voice quality, may serve as early biomarkers for at least some children who will later be diagnosed.13PubMed Central. Early Identification of Autism Using Cry Analysis: A Systematic Review and Meta-analysis of Retrospective and Prospective Studies Some researchers have gone further, proposing that atypical crying is not just a marker but may itself contribute to different interaction patterns between infant and caregiver.14PubMed Central. Cry, Baby, Cry: Expression of Distress As a Biomarker and Modulator in Autism Spectrum Disorder Motor delays during the first year have also been flagged as a potential early clue, and some researchers urge pediatricians not to dismiss the possibility of autism when an infant shows concerning motor behaviors.15PubMed. Early motor delays as diagnostic clues in autism spectrum disorder
The Gut’s Role in the Timeline
An unexpected piece of the early timeline involves the gut microbiome. Researchers comparing stool samples from five-month-old infants at elevated likelihood of autism to those at low likelihood found distinct microbial differences. The elevated-likelihood infants harbored fewer Bifidobacterium species and more Clostridium-related species. Their stool also showed significantly lower levels of GABA, a neurotransmitter with immune-modulating properties. These early microbiome differences were associated with changes in language development tracked from 5 to 36 months.16Translational Psychiatry. Early-life differences in the gut microbiota composition and functionality of infants at elevated likelihood of developing autism spectrum disorder This does not prove that gut bacteria cause autism, but it does suggest the biological picture is more distributed across the body than the brain-only framing implies, and that detectable differences are present well before behavioral signs are obvious.
When Behavior Becomes Noticeable
For most families, the period between 12 and 24 months is when autism first becomes visible in a child’s behavior. The signs that tend to stand out involve social communication, especially joint attention, the back-and-forth sharing of focus on an object or event with another person. Research has found that a child’s ability to initiate joint attention at 8 months and respond to it at 12 months was linked to autism risk at 18 months.17PubMed Central. Joint Attention and Its Relationship with Autism Risk Markers at 18 Months of Age Prospective studies following high-risk toddlers confirm that lower responsiveness to joint attention at 14 months predicted an autism outcome, and that children later diagnosed showed minimal improvement in this skill between 14 and 24 months compared to peers.18PubMed. Response to joint attention in toddlers at risk for autism spectrum disorder: a prospective study
Other signs parents commonly notice around this age include delayed or absent speech, limited interest in other children, repetitive movements, and strong reactions to sensory input. Atypical sensory-based behaviors are a nearly universal feature of autism, though the specific patterns vary widely from child to child.19PubMed Central. Sensory processing in autism: a review of neurophysiologic findings
When Children Lose Skills They Already Had
A significant minority of autistic children follow a pattern that confuses the timeline further: they develop apparently normally for a period and then lose skills they had already acquired. Estimates of how common this regression is vary, with one review putting the figure at roughly a third of young children with autism, mainly involving loss of speech, though nonverbal communication, social engagement, and play skills can also be affected.20PubMed Central. Developmental regression in autism spectrum disorder A Korean study found a lower rate of about 10%, with a typical onset age of 24 months, and noted that children who experienced regression tended to show more severe restricted and repetitive behaviors than those who did not.21PubMed Central. Loss of Acquired Skills: Regression in Young Children With Autism Spectrum Disorders
An important nuance is that most children who demonstrate regression also had earlier, subtler developmental differences that were not recognized at the time.20PubMed Central. Developmental regression in autism spectrum disorder The regression, in other words, is usually not a sudden shift from fully typical development but a more visible phase of something that was already underway. This does not make it less alarming for parents, but it does mean the common narrative of a child being “fine and then suddenly not fine” is usually an oversimplification.
The Gap Between Parental Concern and Diagnosis
Most parents develop concerns about their child’s development between ages 1 and 2, but the average diagnosis does not arrive until ages 3 to 5.22PubMed Central. Parents’ Early Concerns about Their Child with Autism: Relation to Age of Diagnosis A systematic review covering studies from 2012 to 2019 found the overall mean age at diagnosis to be about 60 months, and when the analysis was limited to studies of children under age 10, it was about 43 months.23PubMed. Age at autism spectrum disorder diagnosis: A systematic review and meta-analysis from 2012 to 2019 That gap of two to three years between early concern and formal diagnosis is significant because it overlaps with the period when early intervention appears to have the largest impact.
Several factors predict later diagnosis. Higher IQ, lower severity of autism traits, and lower socioeconomic status are all associated with a later age at first diagnosis.24Journal of Developmental & Behavioral Pediatrics. Age at First Autism Spectrum Disorder Diagnosis This makes intuitive sense: children whose traits are milder or whose cognitive abilities mask their difficulties are harder to identify. Children in families with fewer resources may have less access to specialists or longer wait times. When families notice signs early and act quickly, the wait between noticing and diagnosis shrinks.25PubMed Central. Exploring Factors of Diagnostic Timing Among Black Autistic Youth
Screening tools for toddlers can help, but they are not perfect. A meta-analysis of toddler screening instruments found a pooled sensitivity of about 72%, meaning they correctly flagged roughly seven out of ten children who would eventually be diagnosed, with a high specificity of about 98%, meaning few false alarms. These tools are considered adequate for detecting autism between 14 and 36 months.26PubMed Central. Toddler Screening for Autism Spectrum Disorder: A Meta-Analysis of Diagnostic Accuracy The gap between “adequate” and “reliable enough that no child slips through” explains why universal screening catches most but not all cases.
Why Girls and Women Get Diagnosed Later
Gender is one of the strongest predictors of diagnostic timing. Women and girls are, on average, diagnosed later than their male counterparts despite having similar levels of autistic traits. Research points to camouflaging as a major contributor: the tendency to consciously or unconsciously mask autistic behaviors by mimicking social norms, rehearsing conversation scripts, and suppressing visible stimming. One study found that the relationship between camouflaging and age at diagnosis was significantly stronger for women than for men, and that women who camouflaged heavily were diagnosed at the latest ages.27PubMed. Does camouflaging predict age at autism diagnosis? A comparison of autistic men and women
The problem compounds over time. Women diagnosed in late adolescence or adulthood describe years of “pretending to be normal,” and report that their gender led various professionals to miss the diagnosis entirely. Many experienced conflicts between autistic traits and expectations tied to a traditional feminine identity.28PubMed Central. The Experiences of Late-diagnosed Women with Autism Spectrum Conditions: An Investigation of the Female Autism Phenotype The diagnostic criteria themselves were developed primarily from studying boys, which means the behaviors clinicians are trained to look for may simply be less visible in girls whose presentation differs in style rather than severity.
Racial and Ethnic Differences in Diagnostic Timing
In the United States, the question of whether race affects the age of diagnosis has produced evolving answers. A large study found that by the end of the study period, the median age at diagnosis across racial and ethnic groups fell within a surprisingly narrow range, between roughly 4.4 and 4.9 years. Asian children had the youngest median age at about 4.5 years, while White children had the oldest at about 5 years, with Black and Hispanic children falling between them.29JAMA Network Open. Racial and Ethnic Differences in Rates and Age of Diagnosis of Autism Spectrum Disorder This finding surprised many, since earlier research had consistently found that Black and Hispanic children were diagnosed later and less often. The picture appears to be shifting as awareness and screening improve, but access to diagnostic services still varies enormously depending on geography and insurance.
When Autism Is Not Recognized Until Adulthood
A growing number of people receive their first autism diagnosis as adults. A large registry-based study found that about 69% of men and 61% of women diagnosed with autism in adulthood had no record of any related developmental diagnosis before age 18. The median age of whatever childhood diagnosis they did have was greater than 10 years for most categories, and only about 16% of men and 9% of women had received any relevant diagnosis before age 12.30PubMed Central. Childhood diagnoses in individuals identified as autistics in adulthood These are not people whose autism appeared in adulthood. The condition was present all along; it simply was not identified.
For adults reflecting on their developmental history, research has found that autistic behaviors generally improve over time, particularly in social and communication domains, though restricted and repetitive behaviors tend to be more persistent. In one study of high-IQ adolescents and adults with autism, five of 38 participants who had met full diagnostic criteria at age five no longer met those criteria at their current age, though all five still had substantial impairments.31Journal of the American Academy of Child & Adolescent Psychiatry. Course of Behavioral Change in Autism: A Retrospective Study of High-IQ Adolescents and Adults This means that some adults seeking evaluation may present quite differently from the child they once were, which adds to the difficulty of late-life identification.
Why Earlier Diagnosis Matters for Intervention
The practical urgency behind closing the gap between biological onset and formal diagnosis is intervention. Research consistently shows that the sooner support begins, the greater its impact on the altered brain circuitry of young children, and the better the outcomes.32PubMed Central. The Efficacy of Early Interventions for Children with Autism Spectrum Disorders: A Systematic Review and Meta-Analysis Several comprehensive and targeted treatment models have demonstrated clear benefits for children under three, with improved outcomes documented over periods as long as two years.33PubMed Central. Early Intervention for Children With Autism Spectrum Disorder Under 3 Years of Age: Recommendations for Practice and Research
The brain is most plastic during the first few years of life, which means early behavioral therapies can shape neural pathways during the period when they are most open to change. Waiting for a formal diagnosis before beginning any support is not always necessary; many developmental interventions, like speech therapy and occupational therapy, can start based on clinical concern alone. But a formal diagnosis opens doors to insurance coverage, school-based services, and targeted programs that are otherwise difficult to access. That is why the multi-year lag between early parental concern and official diagnosis remains one of the most consequential problems in developmental pediatrics.34PubMed. First Alarm and Time of Diagnosis in Autism Spectrum Disorders
How Eye-Tracking and EEG Could Change the Timeline
If autism begins prenatally and current screening tools only work reliably around 14 to 36 months, there is an obvious question: can the biological differences detected in the first year be turned into practical screening methods? Researchers are working on it, and two approaches show the most promise.
Eye-tracking protocols that measure how much an infant looks at social versus non-social content have been refined down to two-minute tests that could be run in a pediatrician’s office. One study found moderate correlations between an infant’s preference for geometric patterns over people and scores on the Autism Observation Scale for Infants.35Scientific Reports. Feasibility of a 2-minute eye-tracking protocol to support the early identification of autism The technology is not ready for universal screening yet, but the trend is toward shorter, cheaper, and more portable systems that could supplement the parent-report questionnaires currently in use.
EEG-based approaches, which capture brain electrical activity through scalp sensors, have shown striking results in research settings, including the near-perfect classification accuracy from 3 months mentioned earlier.9Scientific Reports. EEG Analytics for Early Detection of Autism Spectrum Disorder: A data-driven approach However, these studies typically compare infants at very high familial risk to low-risk controls, which is a different task from screening the general population. Moving from high-risk research cohorts to real-world pediatric practice is a major hurdle that involves questions of cost, training, and false-positive rates at scale. Still, the possibility of identifying atypical brain development months or even years before behavioral signs are clear enough for a clinical diagnosis represents a genuine shift in how early intervention could begin.