The majority of studies of temperament in children with autism spectrum disorder (ASD) use scales normed on typical populations. The present study examined a widely used measure of temperament, the Behavioral Style Questionnaire (McDevitt and Carey in Behavioral Styles Questionnaire, Behavioral-Developmental Initiatives Scottsdale, AZ, 1975) to determine whether it contains the temperament traits theorized by its creators. Neither confirmatory nor exploratory factor analysis, using a sample of children with ASD and a population comparison group, identified the theorized nine temperament factors; many items did not strongly load on any of the original factors. A 10 factor solution best described the ASD data and a 9 factor solution best described the typical group's data. There were substantial similarities in the 9 factor solutions, but groups differed from one another enough to question construct similarity for several factors. These results highlight that more basic psychometric research is needed to better understand the BSQ in children with ASD.