The Outcome Questionnaire-45 (OQ) reliably quantifies level of psychological functioning and change during treatment. The three subscales, however, are not well validated. Could alternative scales, based on personality dimensions or other groupings of psychological problems better explain patterns of response? In Study 1, the intended structure and four alternative models were compared using EFA and CFA in random thirds of a community clinic intake sample (N = 1,822). Oblique and bi-level models were compared. Preferred models were tested for stability in samples from later time points. In Study 2, the models were compared in a nonclinical sample (N = 589). Most bi-level models provided adequate fit per standards previously established for the Outcome Questionnaire-45. A seven-factor model of psychological problems provided better fit than any yet reported for this inventory.