Epidemiologic studies of depression have been difficult to interpret because of differing case definitions and variation in diagnostic procedures between studies. We review data from recent epidemiologic studies in which the new research diagnostic techniques were used. We have divided the data into studies of depressive symptoms, bipolar disorder, and nonbipolar depression. An effort is made to integrate the findings of older studies in light of this new classification. Using this classification, there is less variation in epidemiologic rates (point prevalence, incidence, and lifetime risk) than has been noted in previous reviews. Future directions of research are also discussed.