Self-reports of drug use are used extensively in drug use research and in evaluations of drug abuse treatment and human immunodeficiency virus (HIV) prevention interventions. After summarizing recent research into the validity of self-reports in high-risk populations, the article compares the results of validity studies for a sample of patients in methadone maintenance and a sample of criminally involved young adults. The methadone patients more accurately reported their cocaine use; the article explores possible reasons for this. Study findings indicate that the degree of accuracy in self-reports of drug use depends upon the specific research conditions and the characteristics of the populations studied. Accurate estimates of cocaine/crack use among criminally involved, inner-city young male adults cannot rely solely on self-reports; drug use is reported with moderate accuracy to researchers by clients in treatment. Tables, references