The objective of this study was to consider whether trabecular bone score (TBS) improves on areal bone mineral density (aBMD) measurement alone for the prediction of incident fractures in postmenopausal women. METHODS The OPUS study was conducted in ambulatory European women aged above 55years, recruited in 5 centers followed over 6years. For the assessment of the performance of TBS, baseline Hologic scans from 3 centers (Kiel, Paris and Sheffield) were available. Follow-up for incident fractures was available for 1007 women (mean age 65.9±6.9years). We compared the performance of TBS, aBMD, and their combination, by using net reclassification improvement (NRI, primary analysis) and receiver operator characteristic (ROC) c-statistical analysis with ORs and areas under the curves (AUCs) (secondary analyses). RESULTS 82 (8.1%) subjects with incident clinical osteoporotic fractures, and 46 (4.6%) with incident radiographic vertebral fractures were recorded over 6years. Performance of TBS was significantly better than lumbar spine (LS) aBMD for the prediction of incident clinical osteoporotic fractures (NRI=16.3%, p=0.007). For radiographic vertebral fractures, TBS and LS aBMD had similar predictive power but the combination of TBS and LS aBMD increased the performance over LS aBMD alone (NRI=8.6%, p=0.046) but the prediction is similar to hip and femoral neck aBMD. In non osteoporotic women, TBS predicted incident fragility fractures similarly to LS aBMD. CONCLUSIONS This prospective study shows that in general population, TBS is a useful tool to improve the performance of lumbar spine aBMD for vertebral osteoporotic fractures.