Undergraduate General Studies students, following classification by the Bem Sex Role Inventory, formed groups of 44 Stereotyped Masculine males, 36 Androgynous males, 33 Androgynous females, and 25 Stereotyped Feminine females. They predicted achievement on a neutral task and two sex-linked tasks, one of each sex. Almost no variance appears in the Androgynous and Stereotyped predictions for each sex on the neutral task. However, dramatic differences between Androgynous and Stereotyped groups within the same sex do appear in predictions on the sex-linked tasks. We conclude that differences in achievement expectations are less a product of sex differences than of role interpretation differences within each sex. Classroom implications for stereotyping/sex-linking tasks are discussed. (Author)