Background: Psychosocial interventions for substance dependence have demonstrated efficacy. However, the mechanisms by which specific intervention strategies exert their effect have not been clearly identified. Objective: This study investigated the prospective relationships between two psychological processes, an attentional bias toward cocaine stimuli and beliefs about the consequences of cocaine use, and treatment outcome. Method: Twenty-five cocaine-dependent participants enrolled in a 6-month outpatient treatment program that included voucher incentives for abstinence. All participants were asked to complete two implicit assessment procedures, a Drug Stroop protocol and an Implicit Relational Assessment Procedure (IRAP), as well as explicit measures of cocaine craving and the consequences of cocaine use, prior to beginning treatment. Pearson’s correlation coefficients tested the prospective relationships between treatment outcome and the implicit and explicit assessments. Results: Stronger implicit beliefs about the positive effects of cocaine use prior to treatment were associated with poorer treatment outcome when an escalating voucher-incentive program was in place. Further, an attentional bias for cocaine-related stimuli was associated with better treatment outcome when an escalating voucher-incentive program was removed. No association between cocaine use beliefs and treatment outcome was found when beliefs were measured with self-report instruments. Conclusions and scientific significance: These findings highlight the potential utility of performance-based measures for delineating the psychological mechanisms associated with variation in response to treatment for drug dependence.