Genetic association studies of multiple populations investigate a wider range of risk alleles than studies of a single ethnic group. In this study, we developed a multiethnic tagging strategy, exploiting differences in linkage disequilibrium (LD) structure between populations, to comprehensively capture common genetic variation across 60 genes spanning multiple DNA repair pathways, in five racial/ethnic populations. Over 2600 SNPs were genotyped in each population and single- and multi-marker predictors of common alleles were selected to capture the LD patterns specific to each group. Coding variants (n = 211) were genotyped to test whether combinations of putative functional variants in DNA repair pathway genes could have cumulative effects on risk. Tests of association were conducted in a multiethnic breast cancer study (2093 cases and 2303 controls), with validation of the top allelic associations (P </= 0.01) performed in additional studies of 6483 cases and 7309 controls. A variant in the FANCA gene (rs1061646, 0.15-0.68 frequency across populations) was associated with risk in the initial study (P = 0.0052), and in the replication studies (P = 0.032). In a combined analysis (8556 cases and 9605 controls), this SNP yielded an 8% increase in risk per allele. Combinations of coding variants in these genes were not associated with breast cancer and together, these data suggest that common variation in these DNA repair pathway genes are not strongly associated with breast cancer risk. The methods utilized in this study, applied to multiple populations, provide a framework for testing in association studies in diverse populations.