Our research revisits prior work by Neapolitan (2005) on the quality and use of race-specific homicide data. Neapolitan reported that correlations between Black homicide offending rates based on arrest data and rates based on data from the Supplementary Homicide Reports (SHR) for samples of large U.S. cities are only moderately strong. He proposed that, given these findings, the respective rates cannot be regarded as valid indicators of the same concept. We extend Neapolitan's research by estimating regression models to determine the extent to which conclusions about the structural covariates of Black homicide offending rates differ depending on the specific measure of the dependent variable. In addition, we have computed three different Black homicide offending rates with the SHR data: (1) A rate based on single victim/single offender incidents; (2) a rate based on all offenders of known race; and (3) a rate based on the number of Black offenders when the race of offender has been imputed. Our analyses reveal that, consistent with Neapolitan's findings, the correlations between the Black offending rate based on the arrest data and the various SHR-based rates are only moderately strong. In the regression analyses, explained variance is comparatively low in the model with the Black offending rate based on arrest data. However, the regression coefficients do not diverge much across models. Overall, our results suggest that empirical findings and substantive conclusions about city-level covariates of Black offending rates might be less sensitive to the selection of data source than is often assumed. Article by guest on December 3, 2014 hsx.sagepub.com Downloaded from 152 Homicide Studies 18 (2)