This study aims to evaluate the causal effect of sodium-glucose cotransporter 2 (SGLT2) inhibition on bone mineral density (BMD), osteoporosis, and fracture risk using genetics. Two-sample Mendelian randomization (MR) analyses were performed utilizing two sets of genetic variants as instruments (six and two single-nucleotide polymorphisms [SNPs]) associated with SLC5A2 gene expression and glycated hemoglobin A1c levels. Summary statistics of BMD from the Genetic Factors for Osteoporosis consortium (BMD for total body, n = 66,628;​ femoral neck, n = 32,735;​ lumbar spine, n = 28,498;​ forearm, n = 8143) and osteoporosis (6303 cases, 325,717 controls) and 13 types of fracture (≤17,690 cases, ≤328,382 controls) data from the FinnGen study were obtained. One-sample MR and genetic association analyses were conducted in UK Biobank using the individual-level data of heel BMD (n = 256,286) and incident osteoporosis (13,677 cases, 430,262 controls) and fracture (25,806 cases, 407,081 controls). Using six SNPs as the instrument, genetically proxied SGLT2 inhibition showed little evidence of association with BMD of total body, femoral neck, lumbar spine, and forearm (all p ≥ 0.077). Similar results were observed using two SNPs as instruments. Little evidence was found for the SGLT2 inhibition effect on osteoporosis (all p ≥ 0.112) or any 11 major types of fracture (all p ≥ 0.094), except for a nominal significance for fracture of lower leg (p = 0.049) and shoulder and upper arm (p = 0.029). One-sample MR and genetic association analysis showed that both the weighted genetic risk scores constructed from the six and two SNPs were not causally associated with heel BMD, osteoporosis, and fracture (all p ≥ 0.387). Therefore, this study does not support an effect of genetically proxied SGLT2 inhibition on fracture risk.