Direct approaches to attitude measurement make it easy for subjects to give socially desirable responses which may affect the outcome. A regression analysis method called policy capturing was used to study prejudicial attitudes in college students and to obtain estimates of the variance accounted for by predictors associated with sex. Students (N=110) rank-ordered 10 qualities for a professor. Results indicated that 21% of the females and 27% of the males exhibited prejudicial attitudes. Sex, or an interaction with sex, accounted for a wide range of the variance. The findings suggest that attitudes toward sex are more subtle than previously believed. The policy-capturing method can be used to detect the degree of prejudice in a sample when it is not detected by direct measurement, and can also be used individually as a diagnostic instrument. (JAC)