Background: This study tested the theoretical assumption that transformations of parent-child relationships in late childhood and adolescence would differ for boys following different offending trajectories. Methods: Using longitudinal multiinformant data of 503 boys (ages 7-19), we conducted Growth Mixture Modeling to extract offending trajectories. Developmental changes in child reports of parent-child joint activities and relationship quality were examined using Latent Growth Curves. Results: Five offending trajectories were found: non-offenders, moderate childhood offenders, adolescent-limited offenders, serious childhood offenders, and serious persistent offenders. Non-offenders reported high and stable levels of relationship quality between age 10 and 16. Adolescent-limited offenders reported a similarly high relationship quality as non-offenders at ages 7 and 10, but a lower and decreasing relationship quality in adolescence. Compared with non-offenders, serious persistent offenders reported poorer parent-child relationship quality at all ages, and a decreasing relationship quality in adolescence. Serious persistent offenders and adolescent-limited offenders reported similar levels and changes in parent-child relationship quality in adolescence. Although serious persistent offenders reported fewer joint activities at age 10 and 13 than non-offenders, a similar linear decrease in joint activities in early to middle adolescence was found for boys in each trajectory. Conclusion: Developmental changes in parent-child relationship quality differ for different types of offenders. This finding has scientific and practical implications. (Contains 2 tables and 3 figures.)