The purposes of this study were to 1) develop and cross-validate a prediction equation for body density with leisure-time physical activity (LTA) serving as a potential predictor, and 2) determine whether the addition of LTA to previously published equations would improve their prediction. Body density (hydrostatic weighing) was measured in a validation group of 111 women (aged 18-81 y) and was predicted by age, mass, body mass index (BMI), abdominal skinfold thickness, thigh skinfold thickness, and LTA. This equation was cross-validated on an independent group of women (n = 56). Consequently, groups were combined (n = 167) to yield a new equation that also included age, mass, BMI, abdominal skinfold thickness, thigh skinfold thickness, and LTA (R = 0.92, SEE = 7.9 g/L, P < 0.0001). LTA improved the prediction of equations published on women with relatively low body densities, but not on women with higher densities. In conclusion, LTA is an independent predictor of body density in healthy women and its prediction becomes stronger as body density decreases.
