Reliance on any one of the most frequently used measures attempting to identify drug users (self-report techniques, urinalysis, record data, observational data provided by informants or participant-observers, or construct validity measures) has often been inadequate. Because no one measure is accurate, researchers have tended to seek multiple measures for crossvalidation and for cumulatively distinguishing drug users from various measures and the potential for different degrees of reporting accuracy also furthers the use of multiple measures. Researchers conducting the study reported here interviewed 1,889 male arrestees in 6 cities and conducted 1,769 urinalyses among subjects in an attempt to establish the relative validity of urinalysis versus questionnaire data. The urinalyses identified a subgroup of arrestees who tested positive for heroin, barbiturates, amphetamines, methadone, or cocaine. Although the findings indicate a strong correspondence between the urinalysis results and the interview data for heroin users, the opposite is true for the other drugs tested. The lack of correspondence raises serious questions regarding the validity of self-report data on illicit drug usage. From a comparison of arrest charges for the admitted versus the unadmitted drug users, it appears that those who deny drug use are more heavily involved than nonusers in serious crimes against persons. Reliance on any one measure of drug usage would have provided serious underestimates of the number of drug users in the tested population, for the urinalyses identified users who would not admit drug usage, and the interviews identified drug users not identified by urinalysis. Future studies of drug user populations should use multiple measures including urinalysis and interviews. Footnotes, 24 references, and 14 tables are included. For the complete text of the drug use and crime report, see NCJ 40293.