RESUMO
A prediction equation to estimate body fat mass from skinfold thickness for healthy elderly was developed using a four compartmental (4C) model as criterion method. This study included 202 subjects = 60 y old. The measurements of total body water, bone mineral content and body density were included in the 4C model equation. Total sample was randomly partitioned. Sub-sample one was used to design the equations, which were applied in sub-sample two. Its accuracy and precision was evaluated by lineal regression analysis and the bias by Bland and Altman analysis and simple lineal regression. The best model included body mass, sex and the calf and triceps skinfolds thicknesses, with an R2, standard error of the estimate and Cp of 0.85, 3.2 and 3.2, respectively. When the equation was applied in sub-sample two, it was accurate and precise, it showed no significant deviation from the line of identity (the intercept was no significantly different from zero, P>0.05), and slope was different from cero (or similar to 1) (P<0.05). Fat mass by the equation accounted for 86% of the variability of the mean fat mass estimated by the 4C model, having a low standard error of the estimate (3.2 kg) and low pure error (3.1 kg). The new equation was accurate and precise as well as free of significant bias in men and women together and for separately. This equation can be a good option to estimate fat mass in elderly men and women with similar physical characteristics to subjects of this study, and it can be used in clinical and epidemiological studies in this growing group.