The purpose of this study was to investigate the effects of genetic and environmental factors, as well as their interaction, in the etiology of aggressive behavior in two mouse lines bidirectionally selected for offensive aggression. To this end, we raised the Finnish TA (aggressive) and TNA (nonagressive) selection lines either in isolation or in cohabitation with a female after weaning. At the age of 3 months we determined their aggressive behavior in three paradigms (intruder resident, neutral cage, resident intruder) against a male standard opponent. We also determined the animals' aggressive behavior against a female mouse. The results show genetic and environmental effects, as well as gene-environment interaction. We see prominent genotype effects under all conditions but each test is sensitive to a specific combination of environmental effects. A particularly noteworthy result is that variation in the unusual behavior of aggression towards a female is largely explained by the interaction of genotype with isolation. We also examined whether test experience influenced the outcome of an encounter between an experimental animal and an opponent, and found that this factor should not be underestimated, its effect size and direction depending on the type of paradigm and way of housing. These data suggest that the identification of genes underlying aggressive behavior in mice is by no means straightforward and that the result of this search will depend on the environmental design of the study (type of paradigm, housing conditions). These data also suggest that the use of 'test battery' mice might produce different results than the use of test-naive animals.