Objective: To determine whether treatment outcomes are mediated by therapist behaviors consistent with the theoretical postulates on which two contrasting treatments are based. Method: We used data from the U.K. Alcohol Treatment Trial (UKATT), a pragmatic, multicenter, randomized controlled trial comparing the effectiveness of Motivational Enhancement Therapy (MET) and Social Behavior and Network Therapy (SBNT) in the treatment of alcohol problems. N = 376 clients (mean age 42.5, 74.5% male) had 12-month follow-up data and one treatment session recorded and coded using the UKATT Process Rating Scale, a reliable manual-based assessment of treatment fidelity including frequency and quality ratings of treatment-specific therapist tasks and therapist styles. Analyses were conducted using a mediation framework. Results: Analysis of individual paths from treatment condition to treatment process indices (a path) and from treatment process indices to alcohol outcomes (b path) showed that (a) SBNT therapists more often used SBNT-specific behaviors, and did so with overall higher quality; (b) MET therapists more often used MET-specific behaviors, but there was no evidence that they performed these behaviors with higher quality than SBNT therapists; (c) only the quality of MET behaviors significantly predicted 12-month alcohol outcomes, irrespective of treatment condition. Consistently, there were no significant indirect effects. Multiple component analysis indicated that therapist quality of specific tasks influenced outcomes. Conclusions: The quality of delivery of the same treatment tasks in both treatments studied transcended the impact of delivering treatments according to different theoretical underpinnings in UKATT.