Many studies of interventions with high-risk populations have reported reductions in risk behaviors. To assess effectiveness of interventions, data are also needed on the characteristics of subjects lost to follow-up, and on follow-up risk behaviors for subjects who were not participants in the intervention. This paper reports on a study conducted in Harlem, New York, recruiting 1,770 injection drug users (IDUs) and sex partners of IDUs, randomly assigned to two interventions. Repeated-measures analyses for the two intervention groups and those who participated in no intervention indicated that all groups reported significant reductions in risk behaviors, with no group effect. Comparisons of those followed-up and not followed-up indicate that those followed-up were less likely to: be homeless, be Latinos, and to use "shooting galleries." The discussion focuses on the need to assess outcomes for all types of participants, and to distinguish the impact of interventions from other explanations for behavior changes.