The rationale underlying the construction of the Bern Sex Role Inventory (BSRI), and the psychometric properties of the resulting instrument are scrutinized. Following a brief discussion of different types of sex role research, the results of two studies are presented and discussed. In Study 1, 1,464 graduate students rated the desirability of the BSRI traits for one of three referents: man, woman, or adult in American society. In Study 2, 571 graduate students used the BSRI for self-ratings. It was found that, regardless of the referents used, the "masculine" traits were relatively high in desirability but some of the "feminine" traits were low in desirability. Discriminant function analyses revealed that discrimination among groups was primarily due to the differential ratings of the two traits Masculine and Feminine for the different referents. Results from factor analyses of the ratings of desirability and the self-ratings indicated that (a) Bern's classification of the BSRI traits into masculine, feminine, and neutral is not tenable; (b) the dimensions that underlie desirability ratings differ from those that underlie self-ratings; and (c) the dimensions of self-ratings of males differ from those of females. The article concludes with a discussion of the implications of the findings for the measurement of androgyny.Although sex roles have been a subject of study for some time, interest in them has greatly intensified as a direct result of the activities and writings dealing with, or emanating from, the women's liberation movement. (For a review of sex role research, see Hochschild, 1973, and, more recently, Ruble, Frieze, & Parsons, 1976 Investigations of masculinity, femininity, and androgyny have proliferated, emphasizing such aspects as his-
