The validity of generalized and obese population-specific body composition prediction equations for estimating percent fat was examined. Thirty-eight clinically obese women (mean ± SD, % fat = 46.4 ± 5.2%, body mass = 97.2 ± 17.7 kg, age = 41.8 ± 13.3 years) and 16 clinically obese men (% fat = 39.2 ± 3.8%, body mass = 126.7 ± 24.6 kg, age = 43.0 ± 11.6 years) had % body fat estimated by hydrostatic weighing (with RV measured) and had a series of skinfold and girth-based prediction equations applied to estimate % fat. The equations were chosen because they were said to be "generalized" or because they were "population specific" to obese samples of subjects. Results of the present study indicated that the majority of prediction equations were not valid estimators of % body fat in obese males and females. Of the equations that were valid estimators of % body fat (mean differences in % fat of < 3%, r ≥ 0.70, SE < 5%), all but one (Lohman, 1981) utilized girths as part of the prediction technique. It was concluded that generalized girth-based equations could be utilized for prediction of % body fat in obese subjects.