Genetic factors are important in the pathogenesis of osteoporosis but less is known about their possible role in predicting response to anti-osteoporotic therapy. Previous studies have shown that a polymorphic Sp1 binding site in the collagen type 1 alpha 1 gene (COLIA1) is associated with bone mineral density (BMD) and osteoporotic vertebral fracture. In this study we sought to determine if the COLIA1 Sp1 polymorphism might also act as a predictor of the response to treatment of osteoporosis with bisphosphonate therapy. The study group comprised 108 perimenopausal women with osteopenia who had been randomized to receive cyclical etidronate therapy for 2 years with a 1-year treatment-free follow-up as part of a randomized placebo controlled trial. Bone mineral density was measured at the lumbar spine and femoral neck by dual X-ray absorptiometry and genotyping performed on DNA extracted from peripheral blood leukocytes using standard techniques. The distribution of COLIA1 genotypes was similar to that previously reported in Caucasians with 69 (63.9%) "SS" homozygotes, 38 (35.2%) "Ss" heterozygotes, and 1 (0.9%) "ss" homozygote. There was no association between COLIA1 genotype and response of lumbar spine BMD during etidronate treatment or the follow-up phase. The response of femoral neck (FN) BMD, however, differed significantly between the genotype groups throughout the study period, such that FN BMD increased by 0.56%, 2.36%, 1.82%, and 1.32 % after 1, 2, 2.5, and 3 years, respectively in the "SS" genotype group, compared with -1.56%, -0.62%, -0.37%, and -0.66% in the "Ss/ss" genotype groups (P = 0.002). The data presented here show that site-specific heterogeneity exists in the response of BMD to cyclical etidronate therapy, which is related to COLIA1 genotype. Our data raise the possibility that COLIA1 genotyping could be used to target etidronate therapy to those most likely to respond in terms of FN BMD, with potential benefits in terms of economic cost and clinical outcome.