BACKGROUND: Prospective population-based follow-up study of 120 individuals with autism followed from childhood to adulthood. METHODS: Individuals with autism, diagnosed in childhood, were followed prospectively for a period of 13-22 years and re-evaluated at ages 17-40 years. The instruments used at follow-up were the DISCO, WAIS-R, WISC-III, Vineland Adaptive Behavior Scales, psychiatric-medical examination and GAF-scale. A set of criteria was used for the classification of outcomes, taking into consideration employment, higher education/vocational training, independent living and peer relations. RESULTS: Six of the 120 (5%) had died at the time of follow-up, and six declined participation. Overall outcome was poor in 78% of cases. Only four individuals were independent albeit leading fairly isolated lives. Childhood IQ-level was positively correlated with better adult outcome, as was the existence of some communicative phrase speech at age six years. CONCLUSIONS: Children with autism as diagnosed in the 1960s, 1970s, and 1980s may have an even worse psychosocial outcome than previously believed.