Two main findings emerged from the analysis: (1) adult desistance from crime was strongly predicted by both level of self-control and level of social integration; and (2) social bonds predicted desistance from crime regardless of level of self-control. These findings indicate that individuals are likely to desist from offending if they have high self-control or are more socially integrated. However, the second finding shows that informal social control is the key predictor of desistance from offending, regardless of individual background characteristics. The results contradict the social protection hypothesis of life-course interdependence. The finding concerning the importance of social integration in the process of crime desistance underscores the need to develop and sustain social ties among offenders. Data were drawn from a previous longitudinal study of 1,000 Boston males born between 1925 and 1932. The sample included 500 delinquent males recruited from 2 reform schools and 500 nondelinquent males recruited from the Boston public school system. Delinquent participants were matched to nondelinquent participants on age, IQ, and neighborhood socioeconomic status. Data were gathered via interview at ages 14, 25, and 32 and focused on psychological, social, and criminal characteristics. The current analysis focused on desistence from criminal offending using a semiparametric mixed Poisson model (SPMM) that included measures of self-control, social bonds (operationalized as binding life events and social integration), and criminal offending. Future research should attempt to replicate the current findings with a wider range of life events that may contribute to levels of social integration, such as religion and parenthood. Tables, figures, footnotes, references