Most investigations that report a positive association between obstetric complications and schizophrenia have been case-control studies that are often based on long-term maternal recall of events during pregnancy. We tested the hypothesis that mothers of adult offspring with schizophrenia or other psychoses systematically overreport obstetric complications compared with mothers of unaffected offspring. Subjects were selected from the New England cohorts of the National Collaborative Perinatal Project, a large prospective cohort with well-documented records of pregnancy and delivery. Mothers of 39 offspring with psychosis and 39 control offspring were recontacted and completed a structured interview regarding their pregnancy history. Accuracy of maternal recall varied greatly in relation to the type of pregnancy event, and recall was inaccurate for many specific events. For the control sample only, maternal recall of the total number of complications corresponded closely to chart information. Contrary to the study hypothesis, mothers of offspring with psychosis report fewer complications than indicated in their obstetric records, with no evidence of positive recall bias. These results suggest that previous reports of a positive association between obstetric complications and schizophrenia are not likely to have resulted from biased maternal recall.