Background: Dual-energy X-ray absorptiometry (DXA) is being used increasingly to distinguish fat mass (FM) and lean body mass (LBM). However, its accuracy in the paediatric population has only recently received evaluation.

Study design and measurements: Measurements of body composition were made by skinfold thicknesses, DXA (Hologic QDR 1000W) and deuterium dilution, and compared using the Bland Altman method.

Subjects: 42 children aged 5 to 11 years drawn from the general population.

Results: Between-technique correlations were high for LBM (r = 0.95; p<0.001) and FM (r = 0.90; p<0.001). DXA FM also correlated strongly with skinfold thicknesses (r = 0.93; p<0.001) However, mean between-technique difference for both LBM and FM was significantly different to zero. Bland Altman analysis also showed increasing LBM disagreement between DXA and deuterium in smaller and leaner subjects.

Conclusions: This pattern of difference cannot be attributed to the deuterium methodology. DXA provides a good index of fatness but cannot at present distinguish LBM and FM with acceptable accuracy. Such accuracy will require machine- and software-specific calibration against animal carcass analyses.