To determine levels of sustained heroin abstinence, current drug use, and drug-related problems of residential rehabilitation (RR) admissions 24 months after entering treatment. Longitudinal cohort study of 100 heroin users admitted to short-term (1 month) or long-term (3–6 months or longer) RR. Separation in the first week was uncommon in both short-term (7%) and long-term (16%) programs. Eighteen percent successfully graduated, 47% self-discharged, and 30% were expelled. Postindex treatment exposure was widespread (82%), with additional RR the most common treatment. At 24 months, 71% were heroin abstinent over the month preceding interview, and 18% reported heroin abstinence over the entire follow-up period. There had been large declines in levels of recent needle borrowing, crime, psychopathology, and improvements in global and injection-related health. Independent predictors of continuos heroin abstinent were female sex (odds ratio [OR] 5.00), successful graduation from the index program (OR 9.05), and post-treatment MT exposure (OR 0.08). The study confirms the effectiveness of RR and highlights and the impact of program graduation.