This article reviews the operational definition of retirement, which remains an unsettled issue in aging research. The authors describe five separate criteria that can be used to assign retirement status, and discuss the misleading distinction between subjective and objective indicators. Despite concerns about the possible ambiguity of retirement definitions, cross-classifications of persons on three criteria—pension receipt, reduced work or earnings, and self-definition—show considerable overlap depending on the population of interest. Much of the indeterminacy in retirement definitions can be traced to the categorical status of “partial” retirement, which is nevertheless important to recognize in research designs. In all, the ambiguity of retirement definitions can be managed, but not resolved, by using multiple criteria to specify better the retirement status. No one strategy for defining retirement suffices for all purposes; definitions should be suited to the research task at hand.