Abstract
In Baltimore, census tracts made up of White, working-class people are more likely to contain a Toxics Release Inventory (TRI) facility than primarily Black census tracts. Differences in race characteristics decrease with larger units of analysis and with the use of half-mile buffers around TRI sites. At the census-tract level, race is the most significant population characteristic, followed by income and education. A long history of residential and occupational segregation may explain the proximity of toxic-release sites to working-class White neighborhoods.