When the target's gender is the most salient cue, inferences tend to be gender based; when specific behavioral information is more salient, inferences tend to be trait based. This study examined the effects on sex stereotyping of arousal during recall of these two inference situations. Gender- and trait-based inferences were manipulated by providing subjects with information about four targets: One target of each sex was described with four stereotype-neutral behaviors; another target of each sex was described with three stereotype-neutral behaviors plus one stereotype-inconsistent behavior. Subjects memorized the four descriptions under quiet conditions. Later they rated the targets on various gender-related attributes while arousal was manipulated with soft or loud white noise. Results showed that arousal promoted gender inferences for initially gender-based but not for initially trait-based inferences.