Onset and lifetime prevalence of substance abuse were assessed retrospectively using the IRAOS interview in a population-based, controlled sample of 232 first episodes of schizophrenia (ABC sample). Subjects with schizophrenia were twice as likely as controls to have a lifetime history of substance abuse at the age of first admission (alcohol abuse: 23.7 versus 12.3%; drug abuse: 14.2 versus 7.0%). 88% of the patients with drug abuse took cannabis. The sequence of substance abuse and schizophrenia was studied on the timing of abuse onset and illness onset, the latter as based on various definitions: first sign of the disorder, first psychotic symptom and first admission. 62% of the patients with drug abuse and 51% of those with alcohol abuse began the habit before illness onset (=first sign of the disorder). Abuse onset and illness onset occurred highly significantly within the same month (drug abuse in 34.6%, alcohol abuse in 18.2%). Unexpectedly, no temporal correlation was found between abuse onset and the onset of the first psychotic episode. We concluded that a small proportion of schizophrenias might have been precipitated by substance--mainly cannabis--abuse. Long-term effects of early substance abuse were studied prospectively at six cross-sections over five years from first admission on in a subsample of 115 first episodes of schizophrenia. Abusers showed significantly more positive symptoms and a decrease in affective flattening compared with controls. Five-year outcome as based on treatment compliance, utilization of rehabilitative measures and rate of employment was also poorer for patients with than without early substance abuse.