The incidence of diabetes mellitus is rising worldwide, and prediabetic screening for insulin resistance (IR) has become ever more essential. This study aimed to investigate whether body mass index (BMI), waist circumference (WC), or body fat percentage (BF%) could be a better predictor of IR in a middle-aged and elderly population. In this cross-sectional, community-based study, 394 individuals (97 with IR and 297 without IR) were enrolled in the analysis. IR was measured by homeostasis model assessment (HOMA-IR), and subjects with HOMA-IR value ≧75th percentile were defined as being IR. Associations between IR and BMI, WC and BF% were evaluated by t test, chi square, Pearson correlation, logistic regression, and receiver operating characteristic (ROC) curves. A total of 394 community-dwelling, middle-aged, and elderly persons were enrolled; 138 (35%) were male, and 256 were female (65%). The mean age was 64.41 ± 8.46 years. A significant association was identified between BMI, WC, BF%, and IR, with Pearson correlation coefficients of 0.437 (P < .001), 0.412 (P < .001), and 0.361 (P < .001), respectively. Multivariate logistic regression revealed BMI (OR = 1.31; 95% CI = 1.20–1.42), WC (OR = 1.13; 95% CI = 1.08–1.17), and BF% (OR = 1.17; 95% CI = 1.11–1.23) to be independent predictors of IR. The area under curves of BMI and WC, 0.749 and 0.745 respectively, are greater than that of BF% 0.687. BMI and WC were more strongly associated with IR than was BF%. Excess body weight and body fat distribution were more important than total body fat in predicting IR.