The objective of this study was to evaluate the performance of dual energy x-ray absorptiometry (DXA) in forearm measurements of preterm and newborn term infants. The accuracy and linearity of DXA in measuring low mineral content levels (ranging from 30 to 300 mg) was assessed using bone-simulating K2HPO4 phantoms. For in vivo precision, DXA was performed twice on left forearms of four new-born term babies, 21 preterm infants at corrected age 3 mo, and 20 at corrected age 6 mo. Bone mineral content (BMC in mg) and areal bone mineral density (BMD in mg/cm2) at distal forearm and forearm shaft were measured. A special software program allowing a free adjustment of the bone detection threshold was used in the analysis of the scan data. The threshold level affected the overall ability of analysis to detect bone tissue and altered significantly the BMC and BMD values too. Given absolute success in detecting low amounts of bone mineral (BMC > 100 mg), the lowest bone detection threshold evaluated (0.040 g/cm2) became the preferable choice. The relationships between the actual and measured data were highly linear (r was 0.94 for BMC and 0.97 for BMD) but showed underestimation (corresponding slopes were 0.66 and 0.64). In vivo precision expressed as 95% limits of agreement was approximately +/-45 mg for BMC and +/-16 mg/cm2 for BMD. We conclude that DXA provides adequate reliability for in vivo determinations of BMC and areal BMD in the distal and shaft sites of forearm in term and preterm infants and thus strongly supports the clinical utility of DXA in the diagnosis and monitoring of metabolic disease of prematurity. Movements during scanning are typical of pediatric measurements and may decrease the precision considerably. Therefore, every effort must be made to prevent movement artefacts. In addition, special attention must be paid to keeping the analysis procedures consistent.