Despite the success of genome-wide association studies (GWAS) in identifying common genetic variants associated with complex diseases, it has been difficult to demonstrate which variants are causal and their potential contribution to disease. We previously identified via GWAS a variant on 11q13 associated with breast cancer risk (rs614367; OR = 1.21; 95% CI 1.17-1.24; p=10-39). In an attempt to determine the causal variant(s) underlying this association, we analysed 4,405 genotyped and imputed variants in 41 case-control studies within the Breast Cancer Association Consortium (BCAC), which includes 89,050 European subjects. We identified three independent association signals; in each case the SNPs were associated with estrogen receptor (ER)-positive but not ER-negative breast cancer. The strongest signal maps to a transcriptional enhancer element. SNP rs554219, the best candidate causal variant, displays differential binding of transcription factor ELK4, reduces the enhancer activity in luciferase assays and is associated with reduced cyclin D1 levels in tumours. A second, independent SNP, rs78540526, lies in the same enhancer element and also appears to be functional. A third SNP, rs75915166, creates a GATA3 binding site within a silencer element. The effect sizes of the three SNPs are the largest of any GWAS-discovered breast cancer locus identified to date and account for ~2% of the familial risk of breast cancer in Europeans. Finally, we identified, using chromatin conformation studies, a long-range physical interaction restricted to ER+ breast cancer cells between the enhancer and silencer elements and with CCND1, an oncogene with a key role in cell cycle regulation. We conclude that these three variants are likely to be causally related to breast cancer risk and act by controlling cyclin D1 expression.