Common variants in the TERT/CLPTM1L region have been associated with multiple cancers, possibly mediated through control of telomere length (TL). To identify TERT/CLPTM1L variants associated with TL and breast (BC) or ovarian cancer (OC) risk, 110 SNPs were genotyped in 103,991 BC cases/controls, 44,308 OC cases/controls, and 11,705 BRCA1 mutation carriers. TL was determined in blood DNA from 53,645 individuals. We found three independent peaks of association. Peak 1, in the TERT promoter, contained SNPs associated with TL (P=6×10-7), BC risk (P=4.7×10-10), estrogen-negative (ER-) BC risk (P=1×10-9) and BC risk in BRCA1 carriers (P=1×10-5). Luciferase assays showed that a construct carrying the risk alleles of three, highly correlated, candidate functional variants in the TERT promoter abrogated reporter activity, suggesting the risk alleles act by reducing TERT expression. A SNP in Peak 2, which spans TERT introns 2-4, showed the strongest association with TL (P=2×10-14) and risk of serous low malignant potential (LMP) OC (P=1×10-15). SNPs in Peak 3, also spanning TERT introns 2-4, displayed the strongest associations with ER- BC (P=1×10-12), BC risk in BRCA1 carriers (P=1×10-14), and serous invasive OC (P=1×10-11) but were not associated with TL. Luciferase assays showed that the regions around Peaks 2 and 3 can act as silencers and that one of the risk-associated SNPs in Peak 3 decreases relative luciferase signal by ~30%. Analysis of transcripts from a minigene construct showed that one of the risk-associated SNPs in Peak 3 causes expression of a previously unreported TERT mRNA splice variant; analysis of chromatin structure revealed this SNP occupies a novel site of potential regulatory activity in stromal and myoepithelial cells from reduction mammoplasties. Our results show that three independent peaks within TERT are associated with TL, BR and/or OC risk but functional studies indicate that only one of the peaks supports the hypothesis that increased cancer risk is mediated through shorter telomeres.