Large-scale association studies hold substantial promise for unraveling the genetic basis of common human diseases. A wellknown problem with such studies is the presence of undetected population structure, which can lead to both false positive results and failures to detect genuine associations. Here we examine ∼15,000 genome-wide single-nucleotide polymorphisms typed in three population groups to assess the consequences of population structure on the coming generation of association studies. The consequences of population structure on association outcomes increase markedly with sample size. For the size of study needed to detect typical genetic effects in common diseases, even the modest levels of population structure within population groups cannot safely be ignored. We also examine one method for correcting for population structure (Genomic Control). Although it often performs well, it may not correct for structure if too few loci are used and may overcorrect in other settings, leading to substantial loss of power. The results of our analysis can guide the design of large-scale association studies.Recent advances in genotyping technologies and increases in genetic marker availability have paved the way for association studies on genomic scales 1 . A potential problem for every population-based association study is the presence of undetected population structure that can mimic the signal of association and lead to more false positives or to missed real effects (Fig. 1). These concerns have influenced the design, interpretation and funding of association studies during the past decade 2 . Still, levels of population structure in many ethnic groups are typically small, and despite concerns 3,4 , there is an increasing sense 5,6 that the problem is not serious if association studies avoid gross levels of population structure.Upcoming association studies will genotype many markers and evaluate many individuals, owing to the realization that case-control studies powered to detect realistic effect sizes will typically require thousands of individuals 7,8 . This concern raises two general questions: (i) how much underlying structure is there in various human populations and when might this pose problems for large-scale association studies, and (ii) how accurate and efficient are available methods for correcting for population structure in case-control studies?Using genome-wide single-nucleotide polymorphisms (SNPs) in multiple populations (European Americans, African Americans and Asians of known Japanese or Chinese ancestry), we quantified the extent of population structure within and between the populations and then examined the consequences of population structure for association studies. Figure 1 The effects of population structure at a SNP locus. If the study population consists of subpopulations that differ genetically, and if disease prevalence also differs across these subpopulations, then the proportions of cases and controls sampled from each subpopulation will tend to differ, as will allele or genotype fre...