To assess the power of premenstrual changes as a risk factor for future major depressive disorder (MDD), we conducted a follow-up study of 36 women who had volunteered for menstrual cycle studies. Scores on the depressive subscale of the Premenstrual Assessment Form (PAF) at initial evaluation were found to be significantly correlated (r = 0.35) with the occurrence of MDD during the follow-up period. Moreover, multiple regression analysis indicated that the PAF scores had predictive value above and beyond 2 known risk factors for MDD, family history of depression and prior personal history of depression. The Premenstrual Change Index, a score derived from prospective daily self-ratings of severity of dysphoric symptoms, was also correlated with interval MDD, but did not enhance the predictive power of the PAF score. We conclude that the assessment of premenstrual depression has validity in identifying women at risk for future MDD, even when a retrospective instrument, PAF, is utilized for such assessment.