Results indicated that the prevalence and predictors of crime desistance were different with the new measures of desistance. Specifically, relationship quality was consistently related to each crime desistance measure, but the effects of prior crime, peer relationships, race, gender, and parental status varied with the outcome under consideration. The findings indicated the generality of the desistance process and the necessity of comparing accounts of this process with official and self-reported behavioral measures. As such, the authors cannot recommend the adoption of one crime desistance measure over the others. Rather, the specific measure of crime desistance should be based on the theoretical question under examination. To evaluate the new measures of crime desistance proposed in this article, data were drawn from the Youth Development Study (YDS), a longitudinal survey of 1,000 students who attended public schools in Saint Paul, MN during the 1980s. Beginning during their freshmen year in high school, participants provided information about their school, work, family activities, civic participation, and delinquent involvement. In 2002, when most respondents were aged 29 to 30 years, information was collected about crime desistance and transition to adulthood. Distinctions were made between official desistance, measured through arrest records, and behavioral desistance, measured by self-reports. The prediction of desistance through the different measures was analyzed using logistic regression models. Future research should reconsider the role of demographic variables and life course markers, such as family structure and work quality, in predicting the process of crime desistance. Tables, figure, notes, references