Sample sizes varied because of site differences in case flow during the evaluation period. At each site, approximately 75 percent of the sample were men; racial/ethnic representation varied widely across sites. At each site, all cases assigned to TASC were compared to all cases assigned to the control/comparison group, regardless of whether they received any services and regardless of the duration or intensity of services they did receive. Findings show that at four sites, case management had a favorable effect on the frequency of sex while high on drugs. These effects were moderated by users' baseline level of risk behavior or history of related problem behaviors (drug use and criminal conduct). Effects on unprotected vaginal sex emerged only at one site. Services at this site may have differed from services at other sites in ways not detectable in the data. Also, frequency of unprotected sex may be more amenable to change among juveniles, who dominated the sample in this one site, than among adults, who were targeted at all other sites. 5 figures, 5 tables, 34 references, and appended listing of characteristics tested as potential covariates