Boys and girls have different experiences with the same family setting. Power-control theory joins gender differences in control, specifically maternal control of daughters, to the greater proclivity for risk taking and lower perceived sanction threats of boys compared with girls. This study explored the role of dyadic family relationships in the creation of gender differences in delinquency within families, as well as between family types. Analyses were conducted using data from the National Youth Survey. The data were based on a random probability sample of households, using a multistage, cluster sampling design. The analysis examined the first wave of youth aged 11 to 17, and included a parent survey that provided information on parents’ employment statuses, ages of parents, and the number of siblings in the family. The sampling of households permitted the interviewing of multiple children within families; thus, gender differences in delinquency between siblings within families could be examined. Data is tested at both the individual and family levels. The findings were similar across the two sets of analyses. It was concluded that family-level data allowed researchers to devise more methodologically appropriate measures and theoretically informative models than can be constructed with individual-level data. By constructing the measures as comparisons of brothers and sisters on the dependent and key independent variables, researchers were able to address the notion of power-control theory that is within families that these differences were generated. They were able to control for potential sources of “shared environment” characteristics; and perform a more rigorous test of mediating variables on the gender difference in delinquency to determine if these effects varied across household types. The possibility was raised that the family context is composed of a variety of structural and social influences that can produce similarities and differences in the attitudes and behaviors of opposite-sex siblings. 3 tables, 14 footnotes, 53 references, 2 appendices