The study of desistance, the process by which individuals stop offending, is a dynamic field of interest to both academics and policymakers. This chapter reviews the existing theoretical thinking about desistance, and presents a new perspective on the role of identity change in desistance. We begin by verifying empirically that there are in fact people who offend throughout their life-course, and that desistance is not “normative.” In a departure from usual practice, we discuss these models within the framework of formal time series processes. We then present an argument for why identity change is the most promising theoretical direction for criminologists interested in desistance. Finally, we present long-term hazard models as a “new” approach for studying desistance. We close with a challenge to the field to think not just about ways to cause desistance, but also about ways to identify offenders who are in fact at low risk of reoffending (i.e., people who have desisted).