The purpose of this study is to examine the relationships among obesity and hypertension, diabetes mellitus, and hyperlipidemia in elderly women. The anthropometric indicators of obesity, including the body-mass index (BMI), waist circumference (WC), waist-to-hip ratio (WHR), and waist-to-height ratio (WHtR) were used to predict hypertension, diabetes mellitus and hyperlipidermia, and thus to calculate the optimal cut-off point for each indicators of obesity. The sample consisted of women over age of 65, as derived from data of the 2002 National Health ＂Taiwan Three High Prevalence Survey＂ database. Receiver-operating characteristic (ROC) curves were applied to the obesity indicators to ascertain their value in predicting hypertension, diabetes mellitus and hyperlipidemia. The ＂shortest-distance＂ was used to calculate individual cut-off points for each indicators of obesity, and using logistic regression analysis to investigate whether our proposed cutoff points of obesity indicators predict the occurrence of hypertension, diabetes mellitus and hyperlipidemia, compared to reference cutoff points of obesity that are currently used. The results revealed that four obesity indicators can be used to predict hypertension (except WHR) and diabetes mellitus in elderly women, but not hyperlipidemia. Our proposed cutoff points had greater predictability than did the reference cutoff points. The following as ideal cut-off values were demonstrated in elderly women: WHtR＜0.55, WC＜84.9 cm, and BMI＜24.7 kg/m2 to prevent hypertension as well as WHtR＜0.55, WC＜84.1 cm, BMI＜24.4 kg/m2, and WHR ＜0.86 to prevent diabetes mellitus. In conclusion, all four indicators of obesity are not reliable predicators of hyperlipidermia in women over the age of 65, and cut-off values for obesity indicators in predicting hypertension and diabetes mellitus need to be further loosened for the criterion in elderly women.