Prevalent vertebral fractures and baseline bone mineral density (BMD) predict subsequent fracture risk. The objective of this analysis is to examine whether baseline vertebral fracture severity can predict new vertebral and nonvertebral fracture risk. In the randomized, double-blind 3-year Multiple Outcomes of Raloxifene Evaluation (MORE) trial, 7705 postmenopausal women with osteoporosis (low BMD or prevalent vertebral fractures) were randomly assigned to placebo, raloxifene 60 mg/day, or raloxifene 120 mg/day. Post hoc analyses studied the association between baseline fracture severity and new fracture risk in the placebo group and the effects of placebo, raloxifene 60 mg/day, and raloxifene 120 mg/day on new fracture risk in women with the most severe prevalent vertebral fractures (n = 614). Vertebral fracture severity was visually assessed using semiquantitative analysis of radiographs and categorized by estimated decreases in vertebral heights. Reported new nonvertebral fractures were radiographically confirmed. Baseline vertebral fracture severity predicted vertebral and nonvertebral fracture risk at 3 years. In women without prevalent vertebral fractures, 4.3 and 5.5% had new vertebral and nonvertebral fractures, respectively. In women with mild, moderate, and severe prevalent vertebral fractures, 10.5, 23.6, and 38.1% respectively had new vertebral fractures, whereas 7.2, 7.7, and 13.8% respectively experienced new nonvertebral fractures. Number of prevalent vertebral fractures and baseline BMD also predicted vertebral fracture risk, but the severity of prevalent vertebral fractures was the only predictor of nonvertebral fracture risk and remained a significant predictor after adjustment for baseline characteristics, including baseline BMD. In patients with severe baseline vertebral fractures, raloxifene 60 mg/day decreased the risks of new vertebral [RR 0.74 (95% Cl 0.54, 0.99); P = 0.048] and nonvertebral (clavicle, humerus, wrist, pelvis, hip, and leg) fractures [RH 0.53 (95% CI 0.29, 0.99); P = 0.046] at 3 years. To prevent one new fracture at 3 years in women with severe baseline vertebral fractures with raloxifene 60 mg/day, the number needed to treat (NNT) was 10 for vertebral and 18 for nonvertebral fractures. Similar results were observed in women receiving raloxifene 120 mg/day. In summary, baseline vertebral fracture severity was the best independent predictor for new vertebral and nonvertebral fracture risk. Raloxifene decreased new vertebral and nonvertebral fracture risk in the subgroup of women with severe vertebral fractures at baseline. These fractures may reflect architectural deterioration, independent of BMD, leading to increased skeletal fragility.