A subset of transcription factors like Gli2 and Oct1 are bipotential--they can activate or repress the same target, in response to changing signals from upstream genes. Some previous studies implied that the sex-determination protein TRA-1 might also be bipotential; here we confirm this hypothesis by identifying a co-factor, and use it to explore how the structure of a bipotential switch changes during evolution. First, null mutants reveal that C. briggsae TRR-1 is required for spermatogenesis, RNA interference implies that it works as part of the Tip60 Histone Acetyl Transferase complex, and RT-PCR data show that it promotes the expression of Cbr-fog-3, a gene needed for spermatogenesis. Second, epistasis tests reveal that TRR-1 works through TRA-1, both to activate Cbr-fog-3 and to control the sperm/oocyte decision. Since previous studies showed that TRA-1 can repress fog-3 as well, these observations demonstrate that it is bipotential. Third, TRR-1 also regulates the development of the male tail. Since Cbr-tra-2 Cbr-trr-1 double mutants resemble Cbr-tra-1 null mutants, these two regulatory branches control all tra-1 activity. Fourth, striking differences in the relationship between these two branches of the switch have arisen during recent evolution. C. briggsae trr-1 null mutants prevent hermaphrodite spermatogenesis, but not Cbr-fem null mutants, which disrupt the other half of the switch. On the other hand, C. elegans fem null mutants prevent spermatogenesis, but not Cel-trr-1 mutants. However, synthetic interactions confirm that both halves of the switch exist in each species. Thus, the relationship between the two halves of a bipotential switch can shift rapidly during evolution, so that the same phenotype is produce by alternative, complementary mechanisms.