This paper studies individual characteristics of earnings inequality within the population of blacks and whites in the United States over the period 2005-2017. Beyond education and age serving as a proxy for professional experience, applying a new Shapley income decomposition methodology enables us to isolate and measure two discriminative factors in earnings differences: race and gender. We show that these two factors explain a significant share of total earnings inequality, as defined by the Gini index, for all the geographical administrative divisions used. Whatever the division, the share of earnings inequality associated with gender greatly exceeds that of race. While gender earnings inequality has fallen over time, inequality associated with race has tended to increase since 2010 and is stronger in the Southeast of the country. (C) 2020 University of Venice. Published by Elsevier Ltd. All rights reserved.