The cross-sectional area of visceral and subcutaneous fat in the abdomen was measured with T1 weighted spin-echo images acquired with a 1.5-T magnetic resonance (MR) imager. Four axial images centered on L-4 were acquired in each patient. Outline regions of interest (ROIs) were drawn manually for subcutaneous and visceral fat. The subcutaneous fat cross-sectional area was calculated from the ROIs drawn around the outer and inner margins of subcutaneous fat. Several adaptive processing methods were evaluated for measuring fat in the complex structure of the viscera. These methods were compared with an existing MR imaging measurement method for abdominal fat in 18 patients. The adaptive method that uses the valley between the fat and nonfat distributions in the average histogram curve was judged best for research evaluations because it reduces the effects of volume averaging while using a more natural division between fat and nonfat data. Another adaptive method that yielded comparable measurements was thought to be more suitable for clinical applications. Cross-sectional area measurements of abdominal fat were compared in 18 nonobese and 17 obese women to illustrate the utility of these measurements.