The goal of this study is to apply a conventional education cost-function approach for estimating the sensitivity of cost models and predicted education costs to the inclusion of school district level racial composition variables and further to test whether race neutral alternatives sufficiently capture the additional costs associated with school district racial composition. Specifically, our interest is in the role of black student population concentration on the costs of improving commonly measured educational outcomes. Using a three-year panel of data from Arizona, four years from New Jersey, and five from Missouri, we estimate a series of education spending functions and education cost models in order to identify specifically, the marginal cost differences associated with varied racial ethnic student population concentration across school districts. That is, all else equal, does it cost more to achieve any specific level of education outcomes, across school districts by racial composition? We find strong, consistent evidence across settings that black student concentration is associated with higher-predicted costs of achieving constant outcomes, and that those cost differences are quite large for majority black school districts. We discuss briefly the policy and legal implications of these findings. (Contains 8 tables and 21 footnotes.)