The three primary purposes of this investigation were: (1) to apply confirmatory factor analysis (CFA) to responses from three instruments--Bem Sex Role Inventory (BSRI), Comrey Personality Scales (CPS), and Australian Sex-Role Scale (ASRS)--designed to measure masculinity (M) and femininity (F); (2) to determine the correlation between M and F on each instrument and to relate the results to the design of the instrument; and (3) to describe a new theoretical model which posits global M and global F to be multifaceted, higher-order constructs. Responses from two instruments supported the separation of M and F, but differed in the direction of the correlation for the two traits, while the third provided remarkably strong support for a bipolar MF continuum. Despite the apparent inconsistency, the observed correlations were explicable in terms of the design of each instrument. While the two-factor model provided a reasonable fit for the BSRI, more complicated models were better able to fit responses to the ASRS and CPS. Results from this investigation and other research were used to formulate a new theoretical model. (Author/BW)