Surprisingly few studies have investigated the accuracy of prognostic assessments of therapy outcome by clinicians. The objective of the present study was to investigate the relationship between clinicians' prognostic assessments and patient characteristics and treatment outcome. Seventy-one patients with a borderline personality disorder randomly allocated to schema-focused therapy (SFT) or transference-focused psychotherapy (TFP) were assessed every 3 months for 3 years. Prognostic assessments proved to be unrelated to patients' biographical (i.e., age, gender, education level, and employment level) and clinical characteristics (i.e., number of Axis I and Axis II diagnoses, and severity of psychiatric symptoms or borderline personality pathology). Clinical assessors as well as therapists rated the probability of success for SFT to be higher than for TFP. Prospective assessments of assessors and therapists accurately predicted different indices of outcome above and independent of patient characteristics. The prediction of outcome in the TFP condition in particular proved to be valid. Identifying prognostic markers of treatment outcome as used by clinicians in their prognostic assessments may improve current prediction models and patient-treatment matching. © 2008 Wiley Periodicals, Inc. J Clin Psychol: 64: 1–20, 2008.