This study evaluated the accuracy of anthropometric fat location indices (waist circumference and waist-to-hip ratio) as predictors of dyslipidemia, in addition to defining their respective cut-offs. The sample consisted of 416 healthy male blood donors from 20 to 58 years of age at the Blood Center in Cuiabá, capital of Mato Grosso State, Brazil. Serum lipids were determined through the enzymatic method. Data analysis through the ROC curve identified the best indicator and the cut-off with the largest statistically significant area under the ROC curve. The best cut-off points for waist-to-hip ratio and waist circumference were 0.90 and 85cm, respectively. The area under the ROC curve between waist circumference and high cholesterol/HDL relation included the 0.50 value, so waist circumference could not be considered a predictor of dyslipidemia. The best sensitivity and specificity were in the youngest group. Waist-to-hip ratio was a better predictor of dyslipidemia than waist circumference, although the latter showed better discriminating power to detect hypertriglyceridemia. The most accurate cut-off points were lower than those set by the World Health Organization.