Drug use self-reports were compared with urinalysis for 248 clients in four methadone treatment programs. The validity of self-reporting based on urinalysis as a criterion depended on the type of drug examined. Opiate reporting was least valid, while benzodiazepine and cocaine reporting were moderately and highly valid, respectively. EMIT^® urinalysis was far more useful as a criterion of validity than TLC urinalysis. Self-reports helped identify drug users who were missed by urinalysis because of the latter's limited detection period, but urinalysis in turn detected an equal number of drug users missed by the interviews. The age of clients and the type of interviewer directly affected the rate of underreporting. Some respondents systematically denied disvalued behaviors (i.e., drug use and criminality), leading to a spurious correlation between these behaviors. This has important implications for future research.