Most behavior is expressed within social systems, and the genetic analysis of its variance therefore presents theoretical and technical problems that have been sidestepped in most previous research. The dog presents obvious advantages for studying behavioral interactions between genotypes. Two sets of data are summarized that indicate that the magnitude of genetic differences is related to the differentiation of social roles in a competitive situation, whereas similarity of genotypes is an advantage in a situation demanding coordinated activity. Issues concerning the use of correlational techniques to study social behavior are raised.