Baseline data were collected from 1,015 out-of-treatment drug users in Philadelphia as part of an HIV risk-reduction intervention project during a 2.5-year period (March 1992 to August 1994). The study sample was recruited from high drug-using areas in north and south Philadelphia by indigenous outreach workers using targeted sampling. Eligible individuals were at least 18 years old and had either injected drugs and/or used crack and had not participated in drug treatment in the past 30 days. Agreement rates, sensitivity, and specificity measurements were high, and kappa values were good, indicating that out-of- treatment drug users provided moderately valid self-reported drug use. The multivariate analysis showed that women and younger persons were more likely to validly report opiate use, and those who were younger and more educated were more likely to give valid reports of cocaine use. Additional research is needed to better understand differences in the validity of self-reports of opiate and cocaine use and the role that urinalysis plays in influencing the validity of self-reported data. 5 tables, 20 references, and appended formulas used in assessing extent of agreement