The analysis identified a consistent set of predictors that was associated with treatment discharge at 90 days, 12 months, 18 months, and 24 months. Individuals who were HIV seropositive at intake, were younger, used cocaine, drank alcohol daily, and scored high on measures of depression and interpersonal problems were at a higher risk for discharge. Receipt of enhanced methadone treatment -- which included case management services, group participation, psychiatric services, contingency-based reinforcers, and transportation assistance -- was associated with a higher probability of retention, particularly in the first 90 days. These findings can be used to target individuals who are vulnerable to early discharge from treatment and to provide adjunctive services that may improve retention. Several of these predictors, particularly cocaine use and psychological problems, have also been associated with HIV risk among methadone clients. Increasing retention in methadone treatment will not only improve treatment efficacy, but will also address the public health imperative to limit the transmission of HIV. 6 tables, 4 figures, 6 notes, and 45 references