Recent work indicates that marriage contributes to desistance from crime. However, two prominent interpretations of this relationship have been offered. The first, rooted in informal control theory, suggests that the “marriage effect” is a direct result of social bonds that tend to accompany matrimony. The second contends that the effect is indirect and due to the impact of marriage on patterns of delinquent peer association. Using data from waves 5 and 6 of the National Youth Survey, this study re-analyzes these interpretations by examining the relationship between marital attachment, delinquent peer association and desistance from marijuana use. Although change in delinquent peer association is a powerful predictor of marijuana desistance, findings are also consistent with the control theory interpretation of the marriage effect. Implications and limitations of the current study are noted.