In this evaluation of baseline drug use as a predictor of treatment outcome, cocaine use during a 5-week baseline was compared in methadone maintenance patients who had < 5 (n = 10) versus > or = 5 (n = 9) weeks of abstinence during an experimental cocaine abstinence reinforcement treatment. Cocaine use was evaluated at the 1st and last visit and the 1st and last week of baseline and as a mean across the 5-week baseline treatment; response was calculated as a mean across 12 weeks of experimental treatment. Those who had successful outcomes (abstainers) used significantly less cocaine in the 5-week baseline than those with less successful outcomes (nonabstainers). Differences in cocaine use were not evident in the 1st baseline visit or week, but the abstainers used significantly less cocaine in the last visit and week of baseline compared with the nonabstainers. Cocaine use during baseline provided critical predictors of response to the experimental treatment.