Female romantic partners' influence on official crime occurrence for men across a 12-year period in early adulthood was examined within a comprehensive dynamic prediction model including both social learning and social control predictors. We hypothesized that relationship stability, rather than attachment to partner, would be associated with reduced likelihood of crime, whereas women's antisocial behavior would be a risk factor, along with deviant peer association. Models were tested on a sample of at-risk men [the Oregon Youth Study (OYS)] using zero-inflated Poisson (ZIP) modeling predicting to 1) arrest persistence (class and count) and 2) arrest onset class. Findings indicated that women's antisocial behavior was predictive of both onset and persistence of arrests for men, and deviant peer association was predictive of persistence. Relationship stability was protective against persistence.Keywords adult onset; crime in early adulthood; desistance; partner influence; persistence; deviant peer association; attachment to partner Deborah Capaldi is a senior scientist at the Oregon Social Learning Center in Eugene, Oregon. Originally from England, she received her Ph.D. degree in developmental psychology from the University of Oregon. Her research centers on the causes and consequences of antisocial behavior across the life span, including aggression in young couples' relationships, depressive symptoms, health risking sexual behaviors, substance use, and the transmission of risk across three generations. Her recent articles have addressed aspects of all these topics and have appeared in Development and Psychopathology, Journal of Family Psychology, Journal of Marriage and Family, and Journal of Abnormal Psychology. Hyoun Kim is a research scientist at the Oregon Social Learning Center in Eugene, Oregon. Her research interests include individual, social, and contextual influences on the development of psychopathology from adolescence through early adulthood, including depression and other Deborah M. Capaldi, Ph.D., 10 Shelton McMurphey Blvd., Eugene, OR 97401, deborahc@oslc.org, Phone 541-485-2711, FAX 541-485-7087. 1 An alternative model for count data with excess zeros could be provided by using the zero-inflated negative binomial (ZINB) distribution.To check whether an underlying ZINB distribution would provide a better fit to the modeled data than the analyses, assuming an underlying ZIP distribution, the final model containing predictors to class and count was examined within R (R Development Core Team, 2006) using a ZINB distributional assumption. The deviance statistic for the ZINB model was 538.8, compared with the ZIP model with 1 degree of freedom. The resulting difference (chi-square 3.2, 1 d.f.). was not significant at the .05 level, which indicates that the neither the ZIP nor the ZINB model provided a significantly better fit to the data. As an additional test of whether the two-class ZIP model provided an adequate fit to modeling the excess zeros in the Poisson distribution, a parametri...