Vertebral fractures (VFs) are the most common type of osteoporotic fracture, and their prevalence and severity are key risk factors for future fragility fractures. Here, we assess the treatment effect of romosozumab on the incidence of new on-study VFs according to Genant severity grades (mild, moderate, and severe). Data are reported from two phase 3 clinical studies for patients who received romosozumab versus placebo through 12 months, followed by denosumab through 24 months (FRAME:​ NCT01575834), and for patients who received romosozumab through 12 months, followed by alendronate through 24 months, versus alendronate only through 24 months (ARCH:​ NCT01631214). The treatment effect of romosozumab is reported for all included patients, and for patients with prevalent and severe baseline VFs. The incidence of new moderate-or-severe VFs was reduced through 12 months for patients treated with romosozumab versus placebo (FRAME;​ 0.25% versus 1.42%, respectively;​ p < 0.001) or alendronate (ARCH;​ 2.78% versus 4.00%, respectively;​ p = 0.042). Furthermore, the treatment effect of romosozumab on the incidence of new VFs across moderate and severe severity grades was independent of baseline VF prevalence or severity;​ through 12 months, consistent reductions in new moderate-or-severe VFs were observed regardless of prevalent (FRAME;​ p = 0.18) or severe (ARCH;​ p = 0.52) VFs at baseline. Reductions in the incidence of new moderate and severe VFs were sustained through 24 months, after transition from romosozumab to denosumab or alendronate, independent of baseline VF prevalence or severity;​ no significant interactions were observed between the incidence of new moderate-or-severe VFs and the presence of prevalent (FRAME;​ p = 0.81) or severe (ARCH;​ p = 0.99) VFs at baseline. With increasing recommendations for initial treatment with bone-forming agents for postmenopausal women with osteoporosis, these analyses will help to inform treatment decisions for patients at very high risk of VF.