Objective: Epidemiological studies in the United States consistently find autism spectrum disorders (ASD) to be overrepresented in high socioeconomic status (SES) families. These findings starkly contrast with SES gradients of many health conditions, and may result from SES inequalities in access to services. We hypothesized that prenatal measures of low, not high, parental SES would be associated with an increased risk of offspring ASD, once biases in case ascertainment are minimized. Method: We tested this hypothesis in a population-based study in Sweden, a country that has free universal healthcare, routine screening for developmental problems, and thorough protocols for diagnoses of ASD. In a case-control study nested in a total population cohort of children aged 0 to 17 years living in Stockholm County between 2001 and 2007 (N = 589,114), we matched ASD cases (n = 4,709) by age and sex to 10 randomly selected controls. We retrieved parental SES measures collected at time of birth by record linkage. Results: Children of families with lower income, and of parents with manual occupations (OR = 1.4, 95% CI = 1.3-1.6) were at higher risk of ASD. No important relationships with parental education were observed. These associations were present after accounting for parental ages, migration status, parity, psychiatric service use, maternal smoking during pregnancy, and birth characteristics; and regardless of comorbid intellectual disability. Conclusions: Lower, not higher, socioeconomic status was associated with an increased risk of ASD. Studies finding the opposite may be underestimating the burden of ASD in lower SES groups. (Contains 1 figure and 7 tables.)