This paper analyzes the black-white racial gap in US house values in 1998 and 2004 when US house prices experienced considerable appreciation. Un- like previous studies which estimated the racial gap at the mean values, we estimate the entire distribution of the racial di erences in housing wealth. Speci cally, we estimate quantile hedonic regressions for white and black homeowners and then use the Machado and Mata (2005) method to con- struct a counterfactual density of the house values that would have prevailed if houses occupied by whites had the same characteristics as those owned by blacks. This counterfactual density allows us to decompose the racial gap across the whole distribution of house values into a characteristics gap - the part of the gap explained by racial di erences in housing characteristics - and the returns gap - the part of the gap explained by racial di erences in returns to these housing characteristics. Using data from the American Housing Survey, we nd that the racial gap decreased signi cantly at the lower percentiles. In other words, lower-income blacks gained signi cantly relative to lower-income whites from 1998 to 2004. Moreover, the gains can be attributed to a decrease in the returns gap, indicating that either lower- income neighborhoods have become less segregated or that ltering has taken place more e ectively at the lower-income segment of the market. Against this nding, we explore the empirical connection between segregation and ltering and critique the e ectiveness of potential policies and their implica- tions on the distribution of housing wealth.