Neuroblastoma is rarer in African American (AA) children compared to American children of European descent. AA children affected with neuroblastoma, however, more frequently develop the high-risk form of the disease.We have genotyped an AA cohort of 629 neuroblastoma cases (254 high-risk) and 2,990 controls to investigate genetic susceptibility to neuroblastoma in AAs.We confirmed the known neuroblastoma susceptibility gene BARD1 at genome-wide significance in the subset of high-risk cases. We also estimated local admixture across the autosomal genome in the AA cases and controls and detected a signal at 4q31.22 where cases show an increase in European ancestry. A region at 17p13.1 showed increased African ancestry in the subgroup of high-risk cases with respect to intermediate- and low-risk cases. Using results from our published European-American (EA) GWAS, we found that a polygenic score that included all independent SNPs showed a highly significant association (p-value = 1.8 x 10-73) and explained 19% of disease risk variance in an independent EA cohort. In contrast, the best fit polygenic score (p-value = 3.2 x 10-11) in AA included only 22 independent SNPs with association p-value < 2.75 x 10-6 in the EA GWAS, and explained 2% of neuroblastoma risk variance. The significance of the polygenic score dropped rapidly with inclusion of additional SNPs.These findings suggest that several common variants contribute to risk of neuroblastoma in an ancestry specific fashion.This work supports the need for GWAS to be performed in populations of all races and ethnicities.