This study investigates the influence of genetic variation of the estrogen receptor alpha (ESR1) gene locus on several bone parameters in 2042 individuals of The Rotterdam Study, a prospective population-based cohort study of elderly subjects. We analysed three polymorphic sites in the 5' region of the ESR1 gene; a (TA)(n)-repeat in the promoter region, and molecular haplotypes of the PvuII and XbaI RFLPs in intron 1, and inferred long-range haplotypes (LRH) thereof. We observed only three of the possible four PvuII-XbaI haplotypes in our population. A comparison with other Caucasian populations showed similar haplotype frequencies, while in Asian and African populations these were different. Linkage disequilibrium (LD) analysis between the PvuII-XbaI haplotype and the (TA)(n) repeat showed strong LD between the two sites. Reconstruction of long range haplotypes over the entire 5' region, revealed six frequent LRH. In men, we did not observe an association between the ESR1 polymorphisms studied and bone parameters. In women, we demonstrated an allele dose effect of haplotype "px" (P=0.003) and a low number of (TA)(n) repeats (P=0.008) with decreased lumbar spine bone mineral density (BMD) (4.8% lower BMD in women homozygous for haplotype "px", representing 28% of the population, compared with homozygous non-carriers) and decreased vertebral bone area (2.3% difference between extreme genotypes; P=0.016). Most importantly, we found an increased vertebral fracture risk with evidence for an allele dose effect with an odds ratio of 2.2 (95%CI 1.3-3.5) for haplotype "px", and 2.0 (1.5-3.2) for a low number of (TA)(n) repeats. The ESR1 genotype dependent fracture risk is largely independent of BMD and bone area. Combination of risk alleles at both loci by long-range haplotyping improved the associations slightly, but because of the strong LD between the two polymorphic sites, we were unable to determine if any particular polymorphic site is driving the associations found. We conclude that ESR1 polymorphism in the 5' (promoter) region is associated with vertebral fracture risk, lumbar spine BMD and vertebral bone area in postmenopausal women, but not in men. The molecular mechanism underlying this association needs further study.