Bioelectrical impedance analysis (BIA) is a promising tool in the evaluation of body composition in large population studies because it is fast, is inexpensive, and does not require extensive operator training or cross-validation. The empiric nature of the relation between resistance and reactance measured by BIA and body composition has led to the development of equations that translate the raw data into liters of body water or kilograms of fat-free mass (FFM) or fat mass. These equations may not be easily transferred from one population to another if the populations differ significantly in important determinants of body composition such as age, obesity, and illness. I review two recent studies from the Framingham Heart Study in which BIA was first compared with dual-energy X-ray absorptiometry (DXA) as a validation technique, and then compared with the body mass index (BMI, in kg/m2) as an alternative estimate of body fat. BIA was a good predictor of DXA-derived FFM (r = 0.85-0.88, P < 0.001) and was superior to BMI as an estimator of body fat.