The present study was designed to explore the relative efficacy of three types of service delivery intervention models for homeless men with alcohol and/or drug problems: integrated comprehensive residential services provided at one site (Group 1); on-site shelter-based intensive case management with referrals to a community network of services (Group 2); and usual care shelter services with case management (Group 3). In addition to assessing the relative efficacy of these approaches in terms of drug and alcohol use, residential stability, economic and employment status, the project also sought to examine what personal factors best predicted successful outcomes for clients. Clients were assessed at baseline and approximately six months following discharge. All three treatment groups improved significantly over time in terms of reduced alcohol and cocaine use, increased employment, and increased stable housing, but no differential improvement was found among groups. Successful outcomes were predicted by lower recent and lifetime substance use, fewer prior treatment episodes, more stable housing at baseline, fewer incarcerations, and less social isolation.