The high prevalence and serious consequences associated with rape makes its prevention an important social work goal. Rape prevention necessitates understanding the attitudes and personality characteristics of actual and potential rapists. Within the research on attitudinal correlates of rape, hypermasculinity consistently emerges as one of the strongest predictors. The most commonly used measure of hypermasculinity, the Hypermasculinity Index (HMI), uses a forced-choice format that impairs its psychometric properties. This article presents the results of testing a revised version of the HMI using a phrase-completion response format. A convenience sample of undergraduate men (N = 284) from a rural New England university was used. Findings indicate that the new version yields more normally distributed data with a higher internal reliability coefficient. Even more important, the revised version greatly reduced social desirability bias and improved the ability to detect the underlying structure of hypermasculinity. KEY WORDS: hypermasculinity; macho; phrase completion; psychometric properties; rape ********** Sexual assault of adult women by men is a widespread social problem. Lifetime prevalence of sexual assault in the United States is conservatively estimated between 12.65% (Kilpatrick, Acierno, Resnick, Saunders, & Best, 1997; Resnick, Kilpatrick, Dansky, Saunders, & Best, 1993) and 14.8% of women (Tjaden & Thoennes, 2000). Somewhat older and less conservative estimates of lifetime prevalence rates range from approximately one in four (Koss, Gidycz, & Wisniewski, 1987) to one in two (Randall & Haskell, 1995) adult women in the United States. Recent estimates of the annual incidence of rape within the United States range from 95,136 (64.8 per 100,000 women; Federal Bureau of Investigation [FBI], 2002) to 126,500 (103 per 100,000 women; Catalano, 2004). On college campuses there is an average of 35 attempted or completed incidents of sexual assault per 1,000 college women each academic year (Fisher, Cullen, & Turner, 2000). Although the exact number of sexual is unknown as a result of underreporting (Banyard et al., 2005), even these conservative estimates of the prevalence and incidence of sexual assault indicate that it is a widespread social problem. There is evidence that both men and women are capable of being sexually aggressive; however, this article examines only sexual assault of women by men because this focus reflects the reality that men are responsible for the vast majority of sexual assaults (Krahe, Waizenhofer, & Moller, 2003, p. 219). In addition to being widespread, sexual assault also produces serious short- and long-term consequences for the victim (Burgess & Holmstrom, 1974). These consequences include increased anxiety symptoms (Burgess & Holmstrom; Calhoun & Atkinson, 1991), including posttraumatric stress disorder (PTSD) (Calhoun & Resnick, 1993), clinically significant depression (Calhoun & Atkinson; Saunders, Villeponteaus, Lipovsky, Kilpatrick, & Veronen, 1992), sexual issues, and health problems (Calhoun & Atkinson). In addition to these largely individual problems, sexual assault by men can often result in the dissolution of long-term relationships and marriages (Davis, Taylor, & Bench, 1995; Monnier, Resnick, Kilpatrick, & Seals, 2002; Rodkin, Hunt, & Cowan, 1982) as well as lost work and educational productivity, all of which contribute to lowered economic status for rape victims and their children (Calhoun & Atkinson; Monnier et al.). PREDICTING MALE OFFENDERS Given the scope and seriousness of sexual assault by men, the need for sexual assault prevention is obvious and fits well within the purposes of social work (Council on Social Work Education, 2001). However, to prevent rape it is important to improve the ability to predict which men, as a group, are most likely to assault women. …