A growing number of studies show that sexual orientation plays an important role as a predictor of health and social exclusion (e.g., National Research Council, 2011). Consequently, this question gradually became a relevant issue both in public and individual health. The main purpose of this paper is to present a review of current practices from a sociohistorical perspective and to suggest best practices both for a survey design context and for situations of knowledge transfer and use. Before the 19th century, the religio-legal discourse forbidding sodomy focused on the behavioral aspect of sexual orientation. This discourse was replaced by two discourses supporting an identity-based and categorical conceptualization: the emerging medical discourse and the discourse of the first gay-rights movements. During the second half of the 20th century, research and the emergence of the queer discourse questioned the categorical conceptualization and underlined the multidimensional aspect of the construct. The current consensus suggests defining and measuring sexual orientation according to three main components: attractions, behaviors and self-identification (National Research Council, 2011). The proportion of sexual minorities tends to vary according to the specific component being measured, which suggests that these measures target partially exclusive groups. Recent studies showed the importance of systematically taking into account the component of sexual orientation measured in a study and the specific populations that were compared to avoid generalizing conclusions from a component to the other and applying knowledge to the wrong populations. Furthermore, a growing number of studies show the importance of using multiple measures to capture the heterogeneity of the populations identified as sexual minorities. Globally, current best practices in the measurement of sexual orientation mainly underline the need to recognize and assess the diversity of the populations labeled as sexual minorities.