Early onset of antisocial behaviour is one of the most robust predictors of persistent offending across the life course. In broad terms, there are two pathways that follow from this early onset: persistence of extreme antisocial behaviour and desistence from extreme antisocial behaviour. Understanding the pathways away from extreme antisocial behaviour provides useful information in relation to intervention in the lives of those who are at risk for offending. The methodology is guided by Moffitt’s dual typology of life-course persistent and adolescent-limited antisocial behaviour (Moffitt et al 1996). Moffitt’s typological approach also includes a group known as the ‘Recovery’ or ‘low-level chronic’ group (Moffitt 2006). This research focuses on those individuals who exhibited extreme antisocial behaviour in early childhood but by adolescence were no longer classified as extreme: the Recovery group. The data are drawn from the Mater-University Study of Pregnancy (MUSP); a prospective longitudinal study of mothers and their children from the pre-natal period to when the study child was fourteen years of age. The full sample contains 7223 mothers and their children. The MUSP study includes data which allows for the identification of childhood and adolescent antisocial behaviour, as well as a range of variables of key theoretical interest in the study of antisocial behaviour. Moffitt’s typological groupings were replicated using the MUSP data and a series of mean level comparisons between the groups were conducted. Subsequently series of models were estimated to identify the variables which significantly predicted recovery and also to test for gender differences. Findings highlight that those children with an early onset of antisocial behaviour (LCP and Recovery groups) come from the most adverse circumstances when compared to the rest of the sample. However further analyses highlighted that although the Recovery group were faring worse than the majority of the sample in many ways, they were still faring better than the individuals in the LCP group. The models showed that the individual and peer characteristics measured in adolescence were more strongly related to recovery from antisocial behaviour than factors measured earlier in life. Model comparisons also showed that separate models were required for males and females when predicting recovery.