In this post hoc analysis of the Denosumab Fracture Intervention Randomized Placebo-Controlled Trial (DIRECT) in Japanese postmenopausal women and men with osteoporosis, we evaluated the relationship between vertebral fracture risk and both bone mineral density (BMD) T-score and percent change after 24 months of denosumab treatment at total hip, femoral neck, and lumbar spine. Logistic regression analysis was performed and the proportion of treatment effect explained by BMD in vertebral fracture risk was estimated. The results demonstrate that both total hip BMD T-score and change can be strong predictors of subsequent fracture risk, and that total hip BMD change explained 73%, while T-score explained 23%, of the treatment effect. In contrast, neither femoral neck BMD change nor T-score can predict the effect of denosumab on vertebral fracture risk. Furthermore, although lumbar spine BMD T-score was associated with vertebral fracture incidence, lumbar spine BMD change was inversely related to vertebral fracture risk. Because there was no relationship between lumbar spine BMD change and T-score at 24 months of denosumab treatment, and because there can be small undetectable vertebral deformities that may increase BMD values, these results suggest that lumbar spine BMD change is not a good surrogate for vertebral fracture risk assessment. It is suggested that both total hip BMD change and T-score can be good surrogates for predicting vertebral fracture risk in Japanese patients with osteoporosis under denosumab treatment. ClinicalTrials.gov identifier:​ NCT00680953.