Varying measures of sexual orientation are used in women's health research. As they incorporate different dimensions, definitions, and categorical groupings, the comparability of results obtained across studies using different measures remains unknown. We examined the comparability of results using data from the U.S. 2002 National Survey of Family Growth (n = 6,356). Women were classified according to sexual orientation identity, sex of sex partners in the past year, and sex of sex partners over the lifetime. Associations with six health outcomes were compared across sexual orientation schemes. Associations differed in magnitude and statistical significance, even producing conflicting results. Our analyses resulted in a series of methodological recommendations for research on sexual minority women. Data on both behavioral and identity measures should be gathered in health research; identity groups should not be combined for analysis; and researchers should carefully consider which classification scheme(s) to use based on the theoretical basis for the study and the implications for informing interventions.