As HIV-related behavioral research moves increasingly in the direction of seeking to determine predictors of high-risk sexual behavior, more efficient methods of specifying patterns are needed. Two statistical techniques, homogeneity analysis and latent class analysis, useful in scaling binary multivariate data profiles are presented. Both were used to analyze reported sexual behavior patterns in two samples of homosexually active men, one sample of 343 primarily White gay men attending an HIV workshop and one sample of 837 African American gay men recruited nationally. Results support the existence of a single, nonlinear, latent dimension underlying male homosexual behaviors consistent with HIV-related risk taking. Both statistical methods provide an efficient means to optimally scale sexual behavior patterns, a critical outcome variable in HIV-related research.