Women involved in the criminal justice system, particularly those with a history of drug use, are at elevated risk of HIV infection, yet few HIV prevention interventions have been tailored for delivery to incarcerated women. Drawing on the Relational Model, the Reducing Risky Relationships for HIV (RRR-HIV) intervention was developed and evaluated in a multisite randomized clinical trial. Women with weekly drug use prior to incarceration (n = 444) who were incarcerated within correctional institutions in four states were randomized to (1) the RRR-HIV intervention consisting of an HIV educational video, five group sessions, and one postrelease booster session or (2) a control condition consisting of the HIV educational video. The RRR-HIV intervention combined didactic and interactive content regarding seven "thinking myths" about intimate relationships that may result in decisions to engage in risky sexual behaviors. Data were collected while women were still incarcerated and approximately 90 days following release from prison by trained interviewers. A negative binomial regression (NBR) model of unprotected sexual behaviors at the 90-day follow-up indicated that RRR-HIV participants reported fewer unprotected sexual behaviors than women in the control condition once the analysis was adjusted for study site. Future studies should examine the sustainability of the RRR-HIV intervention's effect on risk reduction. Implementation research is needed to determine whether delivery of this intervention by correctional staff or peers, rather than research staff, yields similar reductions in unprotected sexual behaviors.