This article aims to introduce a methodological contribution to the wage decomposition, from the traditional decomposition of Oaxaca–Blinder. Specifically, we review the usual wage decomposition models and proposes a new technique that do not choose an advantage/disadvantage group in advance, a solution to the index problem, one of the most present criticisms in the decomposition salary of Oaxaca–Blinder. It is based on the analysis of the current wages of the analyzed groups, and not on the impacts of the income that the individuals should have, as usually appears in the models of decomposition. It applies the gender wage decomposition procedures in the Brazilian labor market, under the different approaches, from the microdata of the PNAD of 2015, which confirm the wage differences and discrimination between men and women. Finally, this article contributes to the decision on empirical strategies in the field of wage inequalities and, particularly, their interpretation.