A justification for lengthier stays in prison stems from the belief that spending more time in prison reduces recidivism. Extant studies, however, have provided limited evidence for that belief and, indeed, suggest the effect of time served may be minimal. Few studies have employed rigorous methodological approaches, examined time spans of more than one to two years, or investigated the potential for the relationship between recidivism and time served to be curvilinear. Drawing on prior scholarship, this paper identifies three sets of hypotheses about the functional form of the time served and recidivism relationship. Using generalized propensity score analysis to examine data on 90,423 inmates released from Florida prisons, we find three patterns: greater time served initially increases recidivism but then, after approximately one year, decreases it, and, after approximately two years, exerts no effect; estimation of the effects associated with durations of more than five years are uncertain. The results point to potential criminogenic and beneficial effects of time served and underscore the need to identify how varying durations of incarceration affect recidivism.