Sixteen therapists were enrolled in a year-long manualized training program as part of the Vanderbilt II study of time-limited dynamic psychotherapy (TLDP). The training program successfully changed therapists' interventions in line with prescriptions of the TLDP manual, but some unanticipated changes ran counter to the intent of the training, including increased negative interpersonal transactions as indicated by process measures such as the Structural Analysis of Social Behavior (SASB). We examined therapist variables, patient variables, and training variables that appeared to mediate therapist responses to the training program. Results indicate that patient difficulty may mediate certain aspects of therapists' responses to training. Therapists with self-reported hostile and controlling introjects showed the greatest technical adherence, which was intriguing because prior research has linked hostile therapist introject to greater frequency of countertherapeutic interpersonal process. Of special interest were differences in effects of training associated with individual training faculty. This finding, if generalizable, has important implications for manualized therapy research, especially multisite trials.The 5-year Vanderbilt II project examined the effects of 1 year's training in a manualized form of dynamic psychotherapy (time-limited dynamic psychotherapy [TLDP]; Strupp & Binder, 1984) on 16 experienced therapists across 80 cases. Because of the increasing use and acceptance of treatment manuals in psychotherapy research, a primary aim of the study was to explore the process of manualized training itself. A number of therapist behaviors in different domains were measured preand posttraining, with therapists serving as their own controls. Thus, the design focused on the effects of training in a single treatment manual rather than comparing the differential outcome efficacy of several competing protocol therapies.Analyses of therapist changes at the group level revealed both intended and unintended results (Henry, Strupp, Butler, Schacht, & Binder, 1993). The training program successfully changed therapists' technical interventions in line with prescriptions in the TLDP manual. Therapists showed significant increases in interventions, such as focusing the therapeutic process on in-session transactions and the articulation of cyclical maladaptive interpersonal patterns (CMPs). However, other changes were unexpected and ran counter to the intent of the training. These included an increase in negative interpersonal transactions as reflected by such process measures as the Struc-