Using survival methods the authors review and reanalyze nine published reports on relapse after recovery from depression. Despite wide disagreement over cross-sectional rates of relapse, the reanalysis reveals a common finding that the hazard of relapse declines steadily for the first three years after recovery. Methodological and design issues are discussed and summarized for the 40 naturalistic studies that report on longitudinal outcome after recovery from depression. The principles and techniques of survival methods are briefly introduced in an Appendix.