Vitamin D receptor (VOR) polymorphisms have been strongly associated with bone mineral density (BMD) in some studies but not in others. We used a meta-analytic approach to assess quantitatively the association between VDR and BMD and to examine the inftuence of specific study characteristics (e.g., skeletal site, mean age of subjects, menopausal status) on the reported results. Sixteen papers published in peer-reviewedjoumals through July 1996 were included. We calculated the mean dift'erence, percent dift'erence, and effect size (mean dift'erence divided by standard deviation), comparing BMD between homozygous genotypes. At the hip, BMD in the BB genotype was lower than in the bb genotype (mean ditTerence, -0.02 glcm z; percent ditTerence, -2.4%; and effect size -0.18; p = 0.032). At the spine, the mean ditTerencewas -0.03 glcm z; percent dift'erence, -2.5%; and effect size, -0.19; p = 0.062. At the distal radius, the VDR effect was estimated as the mean dift'erence, -0.01 glcm z; percent dift'erence, -1.7%; and etTect size, -0.16; p = 0.078. The spine measurements exhibited the greatest between-and within-study variability. The difference in hip BMD between genotypes was larger (i.e., a more negative number) among the younger women and seemed to decrease with increasing age. However, statistical evidence for this trend was weak (p = 0.(6). Data from the spine and the radius showed no evidence of a comparable interaction of the VDR effect with age. When we omitted data from the first report of an association between VDR polymorphisms and BMD, our analyses gave similar results, although the overall effect estimates were smaller. In the combined data from 29 study groups, the BB genotype frequency was 17.2,4.9, and 2.3% in studies of whites, blacks, and Asians, respectively. VDR polymorphisms represent one genetic factor atTecting BMD, but further research into the mechanisms, clinical significance, and its relation between other genetic and environmental factors is needed. (