Objective: Change talk has been proposed as a mechanism of change in motivational interviewing (MI) by mediating the link between MI technical skills and behavioral outcomes. We tested the influence of therapists' relational skills on this mediation model. Method: Secondary analysis of a randomized controlled trial of individual brief MI for heavy drinking among 20-year-old-Swiss young men, where the MI group (n = 179) significantly reduced drinking compared to an assessment-only control. We coded MI sessions and derived: therapists' MI technical skills, clients' change talk (CT) and sustain talk (ST), and global relational ratings (empathy and MI spirit). We tested moderated mediation models with technical skills as the independent variable, CT and ST as parallel mediators, predicting drinking at 3-month follow-up (controlling for baseline drinking), and relational skills as moderators of the path from technical skills to client mediators. Results: Conditional indirect effects were significant for overall MI technical skills, open questions, and simple reflections (i.e., more of these behaviors related to more ST, which was related to more drinking) when relational skills were low. In contrast, there was a significant conditional indirect effect for complex reflections when relational skills were high (i.e., more complex reflections related to less ST). Conclusions: This study provides partial support for the MI technical and relational process models. Interestingly, support was found regarding the negative side of client ambivalence (ST) in this highly precontemplative sample. Accordingly, MI therapists should work cautiously with ST when clients are at early stages of motivational readiness. (PsycInfo Database Record (c) 2021 APA, all rights reserved).