This study targeted poly-drug (cocaine plus heroin) abstinence among buprenorphine-maintained participants with a 12-week voucher-based reinforcement therapy (VBRT) phase versus a yoked control condition. Baseline levels of cocaine and heroin use were significant predictors of treatment outcome, regardless of treatment assignment. Overall, there were no significant group differences on treatment outcome. However, among the subsample that produced one or more poly-drug-free urine results, VBRT participants had significantly increased cocaine-but not heroin and poly-drug-abstinence, although all results were in the predicted direction. Results suggest that for those who achieve poly-drug abstinence, VBRT may enhance treatment outcome. However, improved interventions, perhaps targeting single-drug abstinence, increasing reinforcement magnitude, or both, may be necessary to promote initial poly-drug abstinence in this population.