The usefulness of the self-schema construct for understanding and predicting human behavior and the reason for the gender-relatedness of certain behaviors and experiences were investigated in three studies. The studies examined cognitive correlates of two gender-related behaviors that are more characteristic of and problematic for women than for men: the avoidance of math and science, and non-participation in athletics. Subjects for Study 1 were female undergraduates at the University of Winnipeg: 184 who had taken more than the required one course in basic science; 213 who had not taken math or science courses; and 49 who were taking physical education courses but no math or science courses. Subjects completed questionnaires measuring their self-schemas for math/science ability, physical ability, and general competence; their perceived physical ability and physical self-presentation confidence; a modified Bem Sex Role Inventory; and several open-ended questions about their performance in courses and athletics. Subjects (N=113) who had been identified from Study 1 data as positive, negative, or aschematic with respect to math/science self-schema or physical/athletic self-schema participated in Study 2. These subjects completed a questionnaire measuring endorsement of self-descriptive items, a generation of behavioral examples task, a recall task, and a math performance test. Male (N=8) and female (N=31) undergraduates in a statistical methods course participated in Study 3 which gathered data on the relationship among self-efficacy, self-schema, and mastery training by using the same self-schema and self-efficacy measures used in Study 1. (Results are presented in detail; a 5-page reference list, 33 data tables, and 2 figures are included.) (NRB)