Multiple skinfold anthropometry (MSA) and bioelectrical impedance analysis (BIA) are useful as clinically non-invasive, inexpensive and portable techniques, although it is not clear if they can be used interchangeably in the same patient to routinely assess her/his body composition. In order to compare BIA, MSA and DXA in the estimation of lean body mass (LBM) of a pediatric obese population, 103 obese [body mass index (BMI) > 97th percentile] children (median age: 11 years; range: 5.4-16.7 years) underwent nutritional evaluation. After an overnight fast, the subjects' anthropometric measurements were performed by the same investigator: body weight (BW), height, skinfold thickness (four sites); fat body mass (FBM) using Brook or Durnin equations and dual X-ray absorptiometry (DXA). BIA was performed using a bioelectrical impedance analyzer (Analicor-Eugedia, 50 kHz) and Houtkooper's equation to calculate LBM. Linear regression analysis was performed to evaluate the relationship between the prediction of LBM by MSA, DXA and BIA. The differences between the three techniques were analysed using Student's t-test for paired observations and the Bland and Altmann method. A considerable lack of agreement was observed between DXA- and BIA-LBM (delta = -4.37 kg LBM; delta-2sigma = -11.6 kg LBM; delta+2sigma = +2.8 kg LBM); between DXA- and MSA-LBM (delta = -1.72 kg LBM; delta-2sigma = -8.2 kg LBM; delta+2sigma = +4.8 kg LBM) and between BIA- and MSA-LBM (delta = -2.65 kg LBM; delta-2sigma = -10.5 kg LBM; delta+2sigma = +5.2 kg LBM). CONCLUSIONS In obese children, DXA, BIA and MSA should not be used interchangeably in the assessment of LBM because of an unacceptable lack of agreement between them. The discrepancies between methods increase with the degree of obesity.