Valid self-assessment is fundamental to continuing professional competence but is seldom explicitly taught in health professions training. This review analyzed 18 scholarly articles published between January 1970 and February 1990 (14 articles regarding health professions trainees, and four concerning college students or graduate trainees) in which it was possible to compare performance as self-assessed by trainees with performance as assessed by experts or objective tests. The validity of self-assessed performance was found to be low to moderate and did not improve with time in conventional health professions training programs. Self-assessed performance seemed closely related to generalized self-attributions and was minimally influenced by external feedback in the form of test scores, grades, or faculty assessments. In five programs emphasizing explicit self-assessment goals and training strategies, moderate-to-high validity outcomes or improvements over time were demonstrated. Much of what passes for self-assessment in training seems the exercise of an underdeveloped skill, but effective training to improve validity and accuracy is available and feasible.