Although it is well established that group cohesion is important for client improvement in group psychotherapy, less is known about how the cohesion—outcome relationship varies over the course of group therapy. No known study has formally tested the association between group cohesion and outcome as a dynamic relationship over the course of group therapy. Moreover, previous research has not routinely monitored cohesion and outcome session-to-session, which has limited researcher’s ability to study the cohesion—outcome relationship at the session and member levels. Therefore, as part of a larger randomized clinical trial evaluating the efficacy of feedback in group therapy, the current study examined the cohesion—outcome relationship over time within-clients and between-clients. More specifically, we partitioned session-to-session cohesion data from 41 members across 5 interpersonal process groups into within-client and between-client predictors of well-being, and modeled interactions with time to test the development of the cohesion—outcome relationship across the span of the groups. Model fit comparisons indicated that a model testing within and between cohesion effects over time was a better fitting model compared to a model that tested within and between components that did not include a time effect. Additionally, our main analyses indicated that time significantly moderated the cohesion—outcome relationship at the client level (between-client effect), but not the session level (within-client effect). This means the between-client cohesion—outcome effect significantly increased over the course of the group. Implications for group practice and future directions are discussed. (PsycINFO Database Record (c) 2019 APA, all rights reserved)