Little empirical research exists about highly effective psychotherapists, and none about the factors that mediate the acquisition and maintenance of superior performance skills (e.g., Ericsson, 1996, 2006; Ericsson, Krampe, & Tesch-Romer, 1993). In the full sample, a 3-level multilevel modeling (level 1: clients; Level 2: therapists; Level 3: organization types) of practitioner outcomes was used to examine the contribution of the therapist to treatment effectiveness. Consistent with prior research, in the full sample (n= 69 therapists; n= 4, 580 clients) it was found that therapist effects explained 5.1% of the variance in outcome, after adjusting for initial severity. Therapist gender, caseload, and age were not found to be significant predictors. In subsample of therapists, the relationship between outcome and therapist demographic variables, professional development activities, and work practices was analyzed (n= 17 therapists, n= 1, 632 clients). Therapist characteristics (e.g., years of experience, gender, age, profession, highest qualification, caseload, degree of theoretical integration) did not significantly predict client-reported outcomes. Consistent with the literature on expertise and expert performance, the amount of time spent targeted at improving therapeutic skills was a significant predictor of client outcomes. Further, highly effective therapists indicated requiring more effort in reviewing therapy recordings alone than did the rest of the cohort. Caveats and implications for clinical practice, continuing professional development, and training are discussed.