Background: Prior twin and adoption studies have demonstrated the importance of both genetic and shared environmental factors in the etiology of criminal behavior (CB). However, despite substantial interest in life-course theories of CB, few genetically informative studies have examined CB in a developmental context. Method: In 69,767 male–male twin pairs and full-sibling pairs with ≤ 2 years’ difference in age, born 1958–1976 and ascertained from the Swedish Twin and Population Registries, we obtained information on all criminal convictions from 1973 to 2011 from the Swedish Crime Register. We fitted a Cholesky structural model, using the OpenMx package, to CB in these pairs over three age periods: 15–19, 20–24, and 25–29. Results: The Cholesky model had two main genetic factors. The first began at ages 15–19 and declined in importance over development. The second started at ages 20–24 and was stable over time. Only one major shared environmental factor was seen, beginning at ages 15–19. Heritability for CB declined from ages 15–29, as did shared environmental effects, although at a slower rate. Conclusions: Genetic risk factors for CB in males are developmentally dynamic, demonstrating both innovation and attenuation. These results are consistent with theories of adolescent-limited and life-course persistent CB subtypes. Heritability for CB did not increase over time as might be predicted from active gene-environmental correlation. However, consistent with expectation, the proportion of variability explained by shared environmental effects declined slightly as individuals aged and moved away from their original homes and neighborhoods.