The latency and topography of attack and flank marking by aggressive male hamsters against olfactory bulbectomized, muzzled, and restrained target hamsters resembled that seen in encounters between unrestrained animals. Subjects' attacks, which occurred ad libitum because the target was restrained, were distributed into bouts of 2-min mean duration. The frequency of agonistic postures and the latency to first attack were significantly affected by castration; these effects were not reversed by exogenous testosterone. The total duration and temporal distribution of attacks were not significantly affected by castration. Attack latency increases and agonistic posture reductions may be a direct effect of castration on aggressiveness, while other changes reported in the literature may reflect the indirect effects of the castrate's reduced aggression-eliciting characteristics on his opponent.