Protein-energy wasting is common in chronic kidney disease and is associated with decreases in body muscle and fat stores and poor outcomes. The accuracy and reliability of field methods to measure body composition is unknown in this population.Cross-sectional observational study.118 maintenance hemodialysis patients were seen at the General Clinical Research Center at Harbor-UCLA Medical Center, Torrance, CA.Triceps skinfold, near-infrared interactance, and bioelectrical impedance analysis using the Segal, Kushner, and Lukaski equations.Dual-energy x-ray absorptiometry (DEXA).Participants (42% women, 52% with diabetes, 40% African Americans, and 38% Hispanics) were aged 49.4 +/- 11.5 (mean +/- SD) years, and had undergone dialysis therapy for 41.1 +/- 32.9 months. Body mass index was 27.0 +/- 6.0 kg/m(2). Using DEXA as the reference test, the bioelectrical impedance analysis-Kushner equation, triceps skinfold, and near-infrared interactance were most accurate of the index tests in estimating total-body fat percentage, whereas bioelectrical impedance analysis-Segal equation and bioelectrical impedance analysis-Lukaski equation overestimated total body fat percentage. Bland-Altman analyses and difference plots showed that bioelectrical impedance analysis-Kushner and near-infrared interactance were most similar to the reference test. Bioelectrical impedance analysis-Kushner, triceps skinfold, and near-infrared interactance had the smallest mean differences from DEXA, especially in women (1.6%, 0.7%, and 1.2%, respectively). Similar results were observed in African American participants (n = 47).Measurements were performed 1 day after a hemodialysis treatment, leading to more fluid retention, which may have affected the reference and index tests differently.Using DEXA as the reference test, both near-infrared interactance and bioelectrical impedance analysis-Kushner method yield more consistent estimates of total body fat percentage in maintenance hemodialysis patients compared with the other index tests. Near-infrared interactance is not affected by skin color. Field methods with portable devices may provide adequate precision.