Bioelectrical impedance analysis (BIA) can potentially be used to estimate body composition in large populations studied at multiple sites. However, it is not clear whether age-specific BIA equations are necessary for accurate application of BIA to research on elderly subjects.We compared a published equation designed to predict fat-free mass (FFM) that had been derived in a young healthy population (mean age 27 y; mean BMI 23.9 kg/m2), with equations that we developed for the elderly by using data from 455 participants in the Framingham Heart Study (78 Y; 27.3 kg/m2), using dual-energy x-ray absorptiometry (DXA) as a reference technique. The BIA equations were then compared in an independent sample of 283 participants in the New Mexico Aging Process Study (76 y, 25.5 kg/m2), who also underwent BIA and DXA.When the young-population equation was applied to Framingham, it caused an overestimation of FFM in heavier subjects that was eliminated by use of the age-specific equation. However, when the two equations were tested in the New Mexico population, the published equation gave estimates of FFM that were closer to DXA than the Framingham equations did.The accuracy of a BIA equation depends on the body composition of the population of the population and the validation method rather than on age per se. Application of BIA to elderly populations requires uniform validation procedures in the actual study population, rather than reliance on age-specific equations.