Longitudinal studies have generally reported poor outcomes in adulthood for the majority of individuals (c.50-60%) with autism. Several factors putatively predict outcome (e.g. IQ), but findings remain mixed. This paper presents an updated meta-analysis of autism outcome studies and extends previous findings with additional analyses (including meta-regression). A total of 4088 records was screened and 18 studies, involving 1199 individuals, were included in the quantitative analysis. Estimated percentages indicated that 20.0% of participants were rated as having a good outcome, 26.6% a fair outcome, and 49.3% a poor outcome. Meta-regression indicated that lower IQ in adulthood was predictive of poor outcome; other meta-regression models did not survive correction for multiple comparisons. Overall, outcomes for autistic people are on average poor, and higher IQ appears to be protective against having a poor outcome. The limitations of current constructs of outcome are discussed.