To identify predictors of treatment outcomes in methadone maintenance programs and to determine whether HIV counseling and testing influenced these outcomes. Retrospective record review. Four methadone maintenance programs in four cities in Connecticut, USA. Five hundred and ninety-four clients, who began treatment over an 18-month period and for whom records were available, took part. HIV counseling and testing. Risk of treatment discontinuation and persistent in-treatment illicit drug use. The most important predictor of treatment discontinuation and of persistent in-treatment illicit drug use was self-reported pre-treatment cocaine use. After controlling for this and demographic risk factors, clients who received initial HIV counseling, when compared with clients who did not, had a similar 12-month discontinuation risk (54 versus 59%; P = 0.08) but were less likely to show persistent illicit drug use (46 versus 53%; P = 0.01). Among counseled entrants who were tested for HIV antibodies, those receiving positive results had a 12-month discontinuation risk similar to those receiving negative results (50 versus 52%), but more often showed persistent illicit drug use (57 versus 44%), although this difference may have been due to chance (P = 0.28). The majority of clients who discontinued treatment did so because they were discharged for non-compliance with clinic rules, usually for failing to pay fees. HIV counseling and testing do not have a substantial adverse effect on methadone treatment outcomes. In the clinics under study, failure to pay clinic fees was an important factor contributing to discontinuation of treatment.