In the traditional model for sexual differentiation in mammals, the female phenotype is the default condition. That is, the female-typical pattern will persist unless acted upon by hormones early in development. The frequency of play fighting in rats, as in most other mammals, is sexually differentiated, and conforms to the traditional model. Males engage in more play fighting than females, and this can be reduced to female-typical levels by neonatal castration. Furthermore, females can be induced to play fight at male-typical levels if treated with testosterone neonatally. Fractionation of play fighting into its constituent components, attack and defense, reveals that it is the frequency of attack that is sexually differentiated, not the likelihood of defense. However, in males, defensive behavior changes at puberty so that the play becomes "rougher." For males to switch to this rougher form of play fighting, they have to be androgenized perinatally. Hence, for males, this second aspect of play fighting that is sexually differentiated also follows the traditional model. In marked contrast, development of the female-typical pattern does not. Neonatal treatment of females with testosterone has no effect; at puberty, they still show the female-typical pattern. On the the other hand, ovariectomy, either at birth or at weaning, leads to females exhibiting the male-typical transition to rougher play fighting at puberty. That is, ovarian hormones appear to actively inhibit the expression of a male-typical trait in females. Play fighting, then, is a mixture of traits, with some features conforming to the traditional model and some not. For some phenotypic features, ovarian hormones appear to exert an active role in their development.