OBJECTIVE Data from a managed care health system were used to examine demographic, clinical, and health system predictors of early dropout from psychotherapy for depression. METHODS A total of 238 members of a mixed-model health plan calling to request an initial psychotherapy visit for depression completed a brief survey regarding depression severity, perceived need for treatment, and expected benefit. Claims records were used to assess therapy visits attended over the following 90 days. RESULTS Dropout rates before the first therapy visit were somewhat higher among women (p=.041) and younger members (p=.017), moderately higher among those with lower depression severity (p=.012), and markedly higher among callers referred to network-model psychotherapists (compared with those given appointments with group-model therapists at health plan clinics) (p<.001). By the second visit, however, cumulative dropout rates were similar for group- and network-model therapists. The only significant predictors of cumulative dropout rates before the second visit were less severe depression at the screening call (p=.004) and lower self-rated importance of initiating psychotherapy (p=.046). CONCLUSIONS Early dropout from psychotherapy for depression was only weakly related to consumers' demographic characteristics. Dropout rates were lower among those with more severe depression, but a significant number of persons with moderate or severe symptoms of depression still discontinued treatment before the second visit. Allocation of visits for individual consumers was markedly different for group- and network-model psychotherapists, but these data did not allow for examination of differences in outcomes for those two treatment models.