Freedman, Matthew L.Pearce, Celeste L.Penney, Kathryn L.Hirschhorn, Joel N.Kolonel, Laurence N.Henderson, Brian E.Altshuler, DavidRepeat length of the CAG microsatellite polymorphism in exon 1 of the androgen receptor (AR) gene has been associated with risk of prostate cancer in humans. This association has been the focus of >20 primary epidemiological publications and multiple review articles, but a consistent and reproducible association has yet to be confirmed. We systematically addressed possible causes of false-negative and false-positive association in >4,000 individuals from a multiethnic, prospective cohort study of prostate cancer, comprehensively studying genetic variation by microsatellite genotyping, direct resequencing of exons in advanced cancer cases, and haplotype analysis across the 180-kb AR genomic locus. These data failed to confirm that common genetic variation in the AR gene locus influences risk of prostate cancer. A systematic approach that assesses both coding and noncoding genetic variation in large and diverse patient samples can help clarify hypotheses about association between genetic variants and disease.Public HealthCancerGenetics and Raceprostate cancerfalse-negativefalse-positivemultiethniccommon genetic variationandrogen receptor (AR) geneArticle