What fraction of US households report having a child with autism, and what do nationally representative surveys show about autism prevalence in children?

The short version: you can ask for a precise household fraction, but nationally representative surveys place autism prevalence in US children at about 2.5–2.8% — roughly one in 36 to one in 40 children.

The most directly useful number comes from the 2016 National Survey of Children's Health (NSCH), a large government-funded survey of over 43,000 US children. It found that 2.79% of children had ever been diagnosed with ASD, and 2.50% had a current diagnosis [3]. That works out to about 1 in 36 children. The same source showed that nearly 30% of those children were receiving neither behavioral therapy nor medication, meaning the diagnosis alone did not guarantee care [3].

An earlier analysis of special education data across 11 years found a different kind of number: autism prevalence among children receiving special education services rose from 1.2 per 1,000 in 2000 to 5.2 per 1,000 in 2010 [2]. That is a 331 percent increase over a decade. The same study found that about 64% of that increase could be accounted for by children who would previously have been classified under intellectual disability instead being reclassified as autistic [2]. That does not mean autism is unreal — it means the diagnostic boundary has shifted heavily over time, and a large part of the apparent epidemic is a relabeling effect.

A much older study from 2003 found a prevalence of 34 per 10,000, or about 0.34% [4]. That shows how far the definition and detection have moved: from 0.3% in the early 2000s to nearly 3% by 2016. This changes lives. At a prevalence of 2.8%, the number of US households with a child diagnosed with autism is in the millions, and the employment and economic burden falls disproportionately on mothers [1].

The evidence in front of me does not give a single national household fraction — how many households have any child with autism — only individual prevalence. The household number would be higher than the child prevalence because some households contain more than one affected child, but this retrieval does not supply it.

My call: the best modern estimate for autism prevalence in US children is about 1 in 36 (2.5–2.8%), solidly established by the NSCH. Confidence: moderate. The lower 0.3% figure from 2003 shows prevalence estimates are highly sensitive to diagnostic criteria and surveillance method, so the modern number is real but reflects broadened criteria and increased awareness, not necessarily an increase in the underlying biology alone.