The degree to which patients and their therapists align over time on their perceptions of therapeutic processes and intermediary outcomes has generally been regarded as an important element of effective psychotherapy; however, few studies have examined empirically the influence of such dyadic convergences on ultimate treatment outcomes. This study examined (a) whether early treatment convergences in patient-therapist alliance and session progress ratings were associated with subsequent worry and distress reduction (and final posttreatment level) in psychotherapy for generalized anxiety disorder (GAD), and (b) whether treatment type and the initial (session 1) levels of perceived alliance and session progress moderated these associations. Data derived from a clinical trial in which patients with severe GAD were randomly assigned to receive either 15 sessions of cognitive-behavioral therapy (CBT; n = 43) or 4 sessions of motivational interviewing (MI) followed by 11 integrative CBT-MI sessions (n = 42). Patients and therapists rated the alliance and session progress after each session. Patients rated their worry and distress multiple times throughout treatment. Dyadic multilevel modeling revealed, as predicted, that alliance convergence over the first half of therapy was associated with greater subsequent worry (p = .03) and distress (p = .01) reduction. Additionally, the combination of low initial patient-rated alliance and early divergence was associated with the worst trajectory for the distress outcome (p = .04). Unexpectedly, session progress divergence had a near significant association with lower posttreatment worry (p = .05) and was significantly associated with more accelerated subsequent distress reduction (p = .03). Additionally, for patients who perceived their initial progress more positively, whether dyads converged or diverged in early session progress ratings affected the trajectories (though not the ultimate amount) of distress change (p = .02). These findings suggest that divergence of early patient-therapist alliance perspectives, especially when coupled with lower initial patient-rated alliance quality, may be an important marker for patient nonresponse and therapist responsiveness (e.g., use of alliance repair strategies). The findings on patient-therapist session progress convergence are less straightforward, though several possible implications are discussed.