Using a sample of males and females at high risk of offending, this study examines associations in criminal behaviour for married partners, over and above pre-existing levels of assortative mating in offending, as well as other individual and family characteristics. Interactions between partner's criminal offending were analysed during first marriage using a longitudinal prospective design. Cox proportional hazard regression analysis for multiple failure-time data showed that both males and females have an immediate increased likelihood of committing an offence when the partner offends. Prolonged associations in offending were observed only for females up until the third month after their partner offended. For both males and females, results showed no long-term associations in partner's offending. © The Author(s) 2013.