Of the three core dimensions of the criminal career--onset, maintenance, and desistance--desistance is the least studied. This study develops a dynamic approach to the measurement and study of desistance and compares it to the more traditional approach. The key defining characteristic of desistance is behavioral change. An ideal measure of desistance should be able to discriminate between people that continue to offend and people that stop offending; estimate whether the change is permanent; and describe the transition from offending to non-offending. Most previous studies of desistance have adopted a static measure of desistance, which counts everyone that offends at least once before a specified cutoff but not afterwards as a desistor. Individuals that offend in both periods are termed persisters. Weaknesses to the static approach are the arbitrary selection of the cutoff point, the heterogeneity of offenders, and the onset of desistance. In the developmental approach, desistance is defined as a process, not a state. This approach emphasizes the transition from offending to non-offending, rather than the state of non-offending. The transition can be characterized by the age range over which the change takes place and the levels of offending that occur during this process. Analysis was conducted with data from the Rochester Youth Development Study, a multi-wave panel study of the development of delinquent behavior among adolescents and young adults. Results show that the dynamic approach has the advantage of being inherently descriptive. A basic picture of the level of behavior at each point in the life course is provided and individuals that have experienced meaningful change in offending can be identified. The chief advantage of this approach--that it provides estimates of rates of offending at different points of time--is also its major drawback. This statistical method is more difficult to implement than the traditional approach. 3 tables, 20 footnotes, 41 references