Studies conducted in the 1970s and early 1980s concluded that people will provide valid information about their illicit drug use when research interviews are conducted under appropriate conditions. Recent studies of treated and untreated populations using improved urinalysis techniques as well as hair analysis techniques indicate that the validity of respondents' self-reports of recent drug use may be considerably less than previously reported and may differ according to a number of factors. Results are presented from a study of clients participating in the Washington, DC, Treatment Initiative study who were assessed for drug use by interview, urinalysis, and hair analysis. At intake, almost all clients who tested positive had reported their use of heroin but fewer clients had reported their cocaine use. At posttreatment followup, clients underreported both heroin and cocaine use. Findings from treatment outcome studies that fail to validate and adjust their estimates of self-reported recent drug use should be interpreted with considerable caution.