When recidivism studies focus on both levels of offending and changes in offending over the lifespan, they also encompass changes that appear to lead to a reduction in recidivism, i.e., desistance from crime. Lifespan analyses that use semiparametric finite mixture models avoid some of the rigidity associated with survival-time models commonly used in recidivism studies. To date, applications of this approach to lifespan data have shown considerable variety in the shape of age-crime trajectories throughout different populations, some of which include a discernible transition from a state of active offending to one of virtual desistance. Trajectory models focus on estimating rates of offending over time and assume that individuals are always active at a rate that changes over time. The use of trajectories of rates of offending to study desistance has brought the study of desistance conceptually close to the study of recidivism. The focus of survival models on the timing of events, on the other hand, may help to resolve the problem of intermittency found in current trajectory analysis. This chapter proposes an approach that recognizes this link and combines the strengths of both approaches in studying desistance. After discussing some of the technical aspects of the survival and trajectory models, the chapter describes a relatively simple statistical model that links them together. 11 notes and 22 references