Survivorship analysis was applied to evaluate treatment responsiveness among four groups of heroin addicts (male and female Whites and Chicanos). Multiple outcome measures included rates of retention, incarceration, addiction, property crime, drug dealing, legal supervision, alcohol abuse, employment, and interpersonal involvement. Differential treatment responsiveness among groups was found depending on the measure in use and the time point of evaluation. Relationships between selected predictors and treatment retention were tested using the Cox proportional hazards model. Young male addicts who were using narcotics daily during treatment and who were unemployed and unmarried were at highest risk for dropping out of treatment.