We reviewed studies measuring unsupervised use of psychoactive substances in schizophrenic and control populations and organized the results by substance class. Despite much variation in their methodologies, these studies broadly agreed that schizophrenic groups' use of amphetamines and cocaine, cannabis, hallucinogens, inhalants, caffeine, and tobacco was significantly greater than or equal to use by control groups consisting of other psychiatric patients or normal subjects. Schizophrenic groups' use of alcohol, opiates, and sedative-hypnotics was significantly less than or equal to use by control groups. We discuss the implications of this nonrandom pattern of drug choice for the hypothesis of substance abuse as a form of self-medication in schizophrenia.