Temporal stability has served as a conceptual basis for the distinction between the clinical syndromes of Axis I disorders and the Axis II personality disorders, the latter being viewed as lifelong enduring patterns. However, comparisons of the stability of Axis I and II disorders have been limited. The present review examines findings from three naturalistic longitudinal studies that utilize similar methodology: the Collaborative Longitudinal Personality Disorders Study (CLPS; Gunderson et al., 2000), the Collaborative Depression Study (CDS; Katz & Klerman, 1979), and the Harvard/Brown Anxiety Research Program (HARP; Keller et al., 1994). Using a definition of remission/recovery as having no or minimal symptoms for 8 consecutive weeks, the courses of personality, depressive, and anxiety disorders were compared. Though remission/recovery rate at the 2-year follow-up was highest for mood disorders, the probability of recurrence was also particularly high. Personality disorders, with remission rates higher than the anxiety disorders, appear to be less stable than conceptualized. The anxiety disorders had remarkably low recovery rates even beyond 5 years of prospective follow-up. Factors that may explain these findings, as well as implications for future conceptualization of DSM, are discussed.