The aim of the present study was to describe and explain the developmental differences in delinquent behaviour from the ages of II to 16 years old between males and females and how family relationships impacted upon the development of delinquent behaviour. The present study was a secondary analysis of the Belfast Youth Development Study, a longitudinal study into adolescent drug use in Northern Ireland, which consisted of approximately 5,000 young people aged 11-12 years old, who began post-primary school in September 2000. Delinquency trajectory groups were estimated using latent class growth modelling for boys and girls separately and multinomial logistic regression models were used to identify a range of family and parenting factors which predicted delinquency trajectory group membership. Mediation analyses were then conducted to assess the mediated effect of parental attachment and parental knowledge variables on the relationship between family structure, family size and delinquency trajectory group. Six delinquency trajectory groups were found for males and four for females. Family structural variables were found to predict delinquency trajectory group membership, as was parental knowledge through parental solicitation, parental control and adolescent disclosure. Parental attachment was a significant predictor for girls but not for boys. However, the effects of family structure and family size were either partially or fully mediated by parental knowledge and attachment in all but one model. In light of these results, implications for policy makers were discussed in relation to gender specific family policies and the design and delivery of delinquency intervention programmes.