Bone strength is predominantly determined by bone density, but bone microarchitecture also plays an important role. We examined whether trabecular bone score (TBS) predicts the risk of vertebral fractures in a Japanese female cohort. Of 1950 randomly selected women aged 15 to 79 years, we analyzed data from 665 women aged 50 years and older, who completed the baseline study and at least one follow-up survey over 10 years, and who had no conditions affecting bone metabolism. Each survey included spinal imaging by dual-energy X-ray absorptiometry (DXA) for vertebral fracture assessment and spine areal bone mineral density (aBMD) measurement. TBS was obtained from spine DXA scans archived in the baseline study. Incident vertebral fracture was determined when vertebral height was reduced by 20% or more and satisfied McCloskey-Kanis criteria or Genant's grade 2 fracture at follow-up. Among eligible women (mean age 64.1 ± 8.1 years), 92 suffered incident vertebral fractures (16.7/10(3) person-years). These women were older with lower aBMD and TBS values relative to those without fractures. The unadjusted odds ratio of vertebral fractures for one standard deviation decrease in TBS was 1.98 (95% confidence interval [CI] 1.56, 2.51) and remained significant (1.64, 95% CI 1.25, 2.15) after adjusting for aBMD. The area under the receiver operating characteristic curve of TBS and aBMD combined was 0.700 for vertebral fracture prediction and was not significantly greater than that of aBMD alone (0.673). However, reclassification improvement measures indicated that TBS and aBMD combined significantly improved risk prediction accuracy compared with aBMD alone. Further inclusion of age and prevalent vertebral deformity in the model improved vertebral fracture prediction, and TBS remained significant in the model. Thus, lower TBS was associated with higher risk of vertebral fracture over 10 years independently of aBMD and clinical risk factors including prevalent vertebral deformity. TBS could effectively improve fracture risk assessment in clinical settings.