Computerized tomography, a method of proven reliability for direct measurement of both total and visceral body fat, is not widely usable and alternative means of assessing body fat are required. Several anthropometric parameters, including ultrasound measurements of intra-abdominal and subcutaneous fat, were considered in 119 females (age 15-72 years, body mass index 24-51 kg/m2). Computerized tomography was also performed for total and visceral fat areas at the L4-L5 level. Reliability of each of these measurements in predicting total fat was assessed. Tertiles for visceral fat area were calculated ( 170 cm2). Only ultrasound intra-abdominal fat measurements were able to distinguish between all three VAT groups (16 +/- 10, 32 +/- 13, 50 +/- 22 mm); neither sagittal diameter nor waist/hip girth ratio were able to distinguish between the two > 114 cm2 groups. Predictive equations for total adipose tissue area confirmed weight/height ratio as the best predictor of total fat. Ultrasound intra-abdominal thickness was the 1 degree variable to enter the stepwise regression in predictive equations for visceral adipose tissue area. The 2 degrees was age, the 3 degrees sagittal diameter, the 4 degrees waist/hip girth ratio and the 5 degrees ultrasound subcutaneous abdominal thickness (adjusted r2 = 0.784; P < 0.001).