Using data from Minnesota for 2000, we show that measures of discrimination in maltreatment substantiation are inflated by a failure to disaggregate counties with large minority populations from those with small minority populations. Racial disparities in substantiation rates, conditional upon reports to child protective service workers, are not huge. Nonetheless, measures of discrimination - once one accounts for characteristics of victims, offenders, reporters, counties and types of maltreatment - are non-trivial. For African Americans, they are higher in the state as a whole than in the counties that have the largest share of minority children. Although the discrimination measures do not vanish when disaggregated analysis is performed, our findings suggest that caution should be displayed when reporting disproportionality statistics that include data from widely dispersed geographical areas.