The Bem Sex Role Inventory (BSRI) is designed to measure masculinity and femininity independently. It is based on the dualistic, as opposed to the bipolar, nature of masculinity and femininity. The development of measures for sex role research has relied on empirical techniques; item selection has been based on respondents' stereotyped conceptions of men and women. This study assessed the extent to which the BSRI traits (representing masculinity, femininity, and a third neutral domain) are actually organized around conceptions of masculinity and femininity. Multidimensional scaling (MDS) was used to identify the perceptual-cognitive structures of sex-typed attributes. Assumptions that the BSRI items are clustered in accordance with conceptions of masculinity and femininity appear unwarranted. Far from being orthogonal, as assumed by the BSRI, the separate "masculinity" and "femininity" dimensions are substantially correlated. Differences observed among the perceptual-cognitive structures of females and males is damaging to the BSRI's validity, as the instrument's identical use (i.e., administration and scoring) with female and male respondents implies highly similar cognitive organization of items for both sexes. It appears that the widespread, uncritical acceptance of the BSRI may well have been premature. (PN)