Whereas estimates of percent deviation of body weight from ideal (F delta weight) are corrected for amputation, those of body mass index (BMI) are not, creating discrepancies in evaluating obesity. A correction of the BMI formula for amputation is proposed. The formula for BMI was corrected for amputation mathematically. The mathematical model predicts that the uncorrected BMI formula underestimates body fat in unilateral amputees and overestimates body fat in subjects with bilateral amputations at the same length of the legs. F delta weight and corrected and uncorrected BMI estimates were computed in 15 subjects with unilateral leg amputation and in 8 subjects with multiple amputations. BMI estimates were as follows: in unilateral amputees, corrected 24.1 +/- 4.1 kg/m2, uncorrected 22.2 +/- 3.9 kg/m2 (p < .001); and in multiple amputees, corrected 21.6 +/- 2.4 kg/m2, uncorrected 32.6 +/- 11.8 kg/m2 (p = .043). Linear regressions of F delta weight obtained from standard nutrition assessment on F delta weight computed from uncorrected and corrected BMI values were as follows: in unilateral amputees, uncorrected F delta weight = -0.079 + 0.932 x actual F delta weight, r = .974, p < .01, and corrected F delta weight = 0.002 + 1.005 x actual F delta weight, r = .997, p < .01; in multiple amputees, uncorrected F delta weight = 0.528 + 1.930 x actual F delta weight, r = .607, p is not significant, and corrected F delta weight = -0.010 + 0.920 x actual F delta weight, r = .936, p < .01.(ABSTRACT TRUNCATED AT 250 WORDS)