It is difficult to measure body fat percentage in clinical settings. Equations using anthropometric measures are more feasible and can be used to estimate body fat. However, there is a need to analyze their accuracy in older adults. Our study aims to validate the use of anthropometric equations to estimate body fat percentage in older men and women. This study evaluated data from 127 Brazilian individuals aged between 60 and 91 years. Weight, height, skinfold thickness and waist and hip circumferences were measured. Seventeen anthropometric equations were tested using the crossed validity criteria suggested by Lohman and the graph analysis proposed by Bland and Altman and by Lin was also performed. The gold-standard method for comparing the anthropometric equations was the dual-energy absorptiometry X-ray (DXA). The average body fat percentage was 30.2 ± 8.6% in men and 43.4 ± 7.9% in women (p < 0.001). In men, the equations which used skinfold thickness presented amplitude of 11.48%, while in women, amplitude's constant error (CE) was 22.88%. The equations based on circumferences and BMI presented CE variation from -5.3% to 29.68% on the estimation of body fat percentage, which means that a same male individual can have the total body adiposity diagnosed with 34.98% of variation, depending on the selection of the employed equation. For women this CE variation was 12.44%. Overall, all the equations yielded different results from the DXA. However, the best equations for male were the one of Lean et al. (1996), which uses the waist circumference, and for women the one of Deurenberg et al. (1991), developed from the body mass index. The need of developing specific equations for older adults still remains, since even the two best equations showed considerable limitations on predicating body fat percentage.