Data was collected through face-to-face interviews in New York City, Fort Lauderdale, Los Angeles, St. Louis, and Phoenix, at four time periods during 1991. The study used two dependent variables, validity of self-reported opiate use and validity of self-reported cocaine use. Multivariate results showed that factors affecting the misrepresentation of drug use varied by site and by drug category. Thus, the validity of self-reports reflected the environment in which interviews occurred and samples on which they were based. Results also suggested the importance of assessing validity in relation to specific drugs, since factors affecting self-report validity often differed for opiate and cocaine users. Explanations for the findings are offered, and ways of improving the validity of arrestee self- reports are discussed. 20 references, 4 notes, and 4 tables