A study examined the effects of gender on social interaction under numerically imbalanced conditions. Specifically, the study tested R. M. Kanter's assumption that all tokens (individuals who enter a work environment where their gender is numerically scarce) respond in a similar manner to token conditions, although evidence exists that males and females respond differently. An interpersonal simulation was run to test both males and females under three different sex ratio conditions. Subjects were 122 librarians--42 males and 80 females. These subjects role-played situations in which they were members of minority groups in the ratios 19:1 (skewed), 6:1 (tilted), and 1:1 (balanced). Effects were measured using the Situational Communication Apprehension Measure (SCAM), because responses in extreme ways would be evidence of perceived performance pressure. The degree of numerical imbalance was found to be linearly related to performance pressure, and males felt significantly less performance pressure than females under the highly skewed sex ratio conditions. Overall, Kanter's theory was supported. (RL)