In this study we compared two parallel self-reported measures that now are being used to assess the recent frequency of intravenous drug use. This study sample consisted of 926 HIV seronegative drug users recruited for participation in HIV research. During a standard interview with each drug user, we first asked about injections in the past 30 days, and then about injections in the past 6 months. The correlation between reports on the past 6 months and the past 30 days was appreciable when all injections were considered (Spearman correlation coefficient rho = 0.78). It increased when the sample was restricted to subjects who reported injections in the past month (rho = 0.88). This restriction resulted in a 15% reduction of the sample size, since 137 participants reported drug use in the previous 6 months but not in the previous 30 days. Concordance tended to be slightly higher for reported frequencies of heroin injections than for cocaine injections, and for men as compared to women. The observed levels of concordance indicate that in many instances both approaches can yield comparable results. Nevertheless the choice of 30 days recall vs 6 months recall must rest upon the specific research questions of each investigation.