Experiences of discrimination are an important aspect of women’s life in Brazil, especially Black women. The Experiences of Discrimination scale (EOD) is often used for assessing discrimination in epidemiological studies, although divergent cutoff points have been used to characterize the exposure. We used latent class analysis (LCA) and logistic regression to identify and characterize subgroups of women exposed to discrimination and compared with a cutoff-based assignment of subgroups. One thousand twohundred and four women living in Salvador, Brazil, responded to the EOD. We selected models with two latent classes, highly and lowly exposed. The classes differed in self-reported skin color and education level, revealing that darker skinned (odds ratio [OR] = 11.3, 95% confidence interval [CI: 1.54, 82.7]) and more educated (OR = 2.09, 95% CI [1.17, 3.72]) women were more likely to be classified into the highly exposed class. Comparing with LCA, the use of cutoff points overestimated the reporting of discrimination. Researchers should consider the use of more accurate measures of discrimination in order to identify the most vulnerable individuals so that prevention efforts could be better targeted.