OBJECTIVES: Using a national database to evaluate the risk of ONJ in osteoporosis females in Taiwan. METHODS: We used the population-based National Health Insurance (NHI) database in Taiwan to conduct a retrospective cohort study during the period from 2000 through 2008. The study population confined to the women who took alendronate or raloxifene initially from Jan, 2001 to Sep, 2005 and medication needed to be stably prescribed for 90 days. The measurements of clinical outcomes included ONJ-diagnosis (ICD-9-CM diagnosis code 526.4, 526.5, 528.3, and 730.18) and sequestrectomy (NHI payment code: 92025B and 92026B) within 3 years after stable medication use. Cox’s proportional hazard model was used to analyze the risk of ONJ development. RESULTS: We identified 23,158 women who had been prescribed either alendronate (n = 18,504) or raloxifene (n = 4,654). The patients treated with alendronate were at a higher risk of ONJ-diagnosis compared with the raloxifene group (hazard ratio [HR] = 1.336 [95% CI, 1.100-1.622]). For sequestrectomy, patients took alendronate also had a higher risk than raloxifene group (HR=4.648 [95% CI, 1.120-19.291]). CONCLUSIONS: Patients who took alendronate had a higher risk of ONJ than raloxifene patients, no matter from the perspective of having the diagnosis or performing sequestrectomy. When undertaking the bisphophonates treatment, doctors need to pay attention to the oral condition of patients.