The data came from a well-known 1972 survey of random samples of the populations of Iowa, Oregon, and New Jersey and from interviews in 1993 with respondents from 57 percent of the originally selected households. The resulting sample compared favorably with census descriptions of the populations of the three States in 1970. Results revealed some nonchance interaction between age and demographic and theoretical predictors of criminal behavior that is localized in specific age categories and around particular variables, offenses, or both. Overall, however, such interaction did not appear to be substantial. In addition, it did not seem to have important consequences for generalizing from age-restricted samples, particularly where generalization is of the most common type: from youth samples to adults. Therefore, despite some results contrary to a strict assertion that the causes and correlates of crime are the same for all ages, the findings support the thrust of the Hirschi-Gottfredson interaction hypothesis. Figures, tables, and 75 references (Author summary modified)