OBJECTIVE: This study investigated two proposed change mechanisms in preference accommodation, thought to improve psychotherapy outcomes: a direct effect of a match between clients' initial preferences and their subsequent experiences of the therapy activities, or a mediated "matching effect" operating through the working alliance. Furthermore, the study explored whether the effect of a preference-experience (mis)match depended on the phase of therapy or type of therapy activity. METHOD: Three hundred sixty-six adults (Mage = 43.2, 75% female) were seen by 50 therapists in individual psychotherapy. Cross-lagged associations between latent preference-activity discrepancies (measured by the Preference and Experience Questionnaire), working alliance (measured by the Session Alliance Inventory), and affective symptoms (measured by the Symptom Checklist-11) were analyzed using dynamic panel modeling adjusted for between-person differences. RESULTS: Generally, the clients' initial preference levels exceeded the amount of therapy activities they experienced receiving. Deviations from this general discrepancy, so that the amount of activities increased in the direction of the client's preference levels, were significantly associated with stronger alliance ratings for all activity types and across most sessions. However, no fully mediated sequence between a change in preference-experience-discrepancy, the alliance, and symptoms was found. Moreover, the direct associations between preference-experience-discrepancy and subsequent symptom change were inconsistent and indicated differences between distinct activity types and phases of therapy. CONCLUSIONS: Both direct effects and alliance effects in preference accommodation were found, but the study provides particular support for matching effects impacting the working alliance and highlights the potential of preference work in alliance development. (PsycInfo Database Record (c) 2025 APA, all rights reserved).