Body composition analysis is an important component of nutritional assessment in cystic fibrosis (CF). No gold standard of measurement exists, and techniques applicable to healthy populations may be unsuitable for CF patients. We assessed lean body mass (LBM) in 12 children with CF by skinfold (SK) measurements, bioelectrical impedance analysis (BIA), and dual-photon absorptiometry (DPA) and repeated these measures in 10 subjects 6 mo later. SK and DPA measures in eight older CF patients and eight healthy controls were compared to evaluate any effect of disease on estimates of LBM by use of DPA. Good agreement between the measures was seen at baseline and 6 mo by use of concordance plots. However, the limits of agreement between measures ranged up to 19% of SK-derived LBM measures (baseline: SK and DPA, 2.63 to -3.93 kg; SK and BIA, 2.36 to -1.24 kg; BIA and DPA, 1.88 to -4.28 kg; 6 mo: SK and DPA, 2.10 to -3.58 kg; SK and BIA, 6.28 to -5.49 kg; BIA and DPA, 5.53 to -7.79 kg). The change in LBM over 6 mo did not correlate among the three measures. Only BIA change in LBM correlated with weight change (r = 0.716, P < 0.02), probably due to the inclusion of weight in the regression equations for determining LBM from impedance. The relationship between SK and DPA measures did not differ between the CF and control groups, suggesting that there was no effect of disease on the DPA measure. The results suggest that none of these methods is precise enough to follow short-term changes in the nutritional status of CF patients longitudinally.