As part of a New Deal initiative to minimize home foreclosure, federal government officials and local real estate professionals graded each neighborhood in America's largest cities on its perceived credit risk. Using recently digitized maps that precisely show neighborhoods marked with red ink (highest risk) or yellow ink (slightly lower risk), I document that surveyors disproportionately assigned the most restrictive credit rating to neighborhoods with black residents. Nearly 90 percent of African Americans in 1940 lived in a census tract marked for credit redlining. Comparing credit-restricted "redlined" census tracts to adjacent "yellow-lined" tracts, I estimate the long-run effects of redlining on housing and neighborhood outcomes. Between 1940 and 1970, redlining was associated with large differential declines in housing supply and population density; homeownership rates and racial composition did not change differentially from their 1940 baseline though. Once discriminatory lending was outlawed during the mid-1970s, there was moderate convergence in homeownership rates and racial composition. However, housing supply and population density remain persistently lower in formerly credit-restricted census tracts relative to their credit-favored neighbors.