Previous research has shown that invariant facial features—for example, sex—and variant facial features—for example, emotional expressions—interact during face categorization. The nature of this interaction is a matter of dispute, however, and has been reported as either asymmetrical, such that sex cues influence emotion perception but emotional expressions do not affect the perception of sex, or symmetrical, such that sex and emotion cues each reciprocally influence the categorization of the other. In the present research, we identified stimulus set size as the critical factor leading to this disparity. Using faces drawn from different databases, in two separate experiments we replicated the finding of a symmetrical interaction between face sex and emotional expression when larger sets of posers were used. Using a subset of four posers, in the same setups, however, did not provide evidence for a symmetrical interaction, which is also consistent with prior research. This pattern of results suggests that different strategies may be used to categorize aspects of faces that are encountered repeatedly.