目標：比較台灣中老年人身體質量指數(BMI)及腰圍與糖尿病的關聯，及其預測罹患糖尿病風險的能力。方法：分析「台灣地區中老年身心社會生活健康狀況長期追蹤調查」2003及2007年4987位≥50歲個案的資料。以邏輯斯迴歸分析BMI和腰圍與糖尿病風險的橫斷關聯，及預測2007年新罹患糖尿病的風險。又以受試者工作特徵(Receiver Operating Characteristics, ROC)曲線計算兩指標的預測切點。結果：兩指標都與糖尿病的風險呈顯著關聯，且都具預測未來罹患糖尿病的功能，但是腰圍比BMI關聯強，預測力也較強。腰圍每增加1 cm，男性增加6%風險(p=0.001)；女性增加4% (p=0.005)。而兩指標判定糖尿病風險的切點因年齡及性別而異，從50-64歲到≥75歲，男性由BMI 24.52 kg/m2減到21.82 kg/m2；女性由24.27 kg/m2減到22.88 kg/m2；而腰圍男性由84.5 cm增至89.5 cm；女性由82.5 cm增至91.5 cm。結論：BMI和腰圍皆具預測中老年人罹患糖尿病的能力，但腰圍預測力較強。而兩指標預測罹患糖尿病風險的切點則因性別、年齡而不同。此結果與現行台灣中老年人糖尿病風險的建議切點有所不同。國人的理想切點值應考量性別及年齡的差別。

Objectives: To evaluate the association of Body Mass Index (BMI) and waist circumference (WC) with the risk for diabetes, and to examine the ability of these measures to predict new cases of diabetes in ≥50-year old Taiwanese. Methods: We utilized the 2003 and 2007 "Survey of Health and Living Status of the Elderly in Taiwan" datasets. We analyzed the cross-sectional associations in 4987 subjects (≥50 years old) and the ability of BMI and WC to predict new cases of diabetes in these subjects four years later. We also analyzed the cutoffs with Receiver Operating Characteristics (ROC). Results: Both indicators were associated with a risk for diabetes but WC showed a stronger association. Both indicators significantly predicted the risk for new diabetes, but WC was a stronger predictor. For each 1 cm increment in WC, there was a 6% increase in the risk for new diabetes in men (p=0.001) and a 4% increase in women (p=0.005). The cutoffs for the risk of new onset diabetes differed by age and gender. The BMI cutoff was reduced from 24.52 to 21.82 kg/m2 for men and 24.27 to 22.88 kg/m2 for women, whereas the WC cutoff was increased from 84.5 to 89.5 cm for men and 82.5 to 91.5 cm for women. Conclusions: Both BMI and WC were significantly associated with the risk for diabetes or the risk for new diabetes, but the association was stronger with WC for both. The predicted cutoffs for diabetes varied according to age and gender and were different from the current Bureau of Health Promotion recommendations.