Dual-energy absorptiometry (DXA) is widely used for bone mineral density measurements. Different types of devices are available. Differences between devices from either the same manufacturer or different manufacturers can lead to difficulties in clinical practice when patients are followed on different machines. We calculated the accuracy and precision of 62 DXA devices from two manufacturers (51 Hologic, 11 Lunar) using a European Spine Phantom (ESP, semi-anthropomorphic). The ESP was measured 5 times on each device without repositioning. Accuracy was assessed by comparing bone mineral density (BMD, g/cm(2)) values measured on each device with the actual value of the phantom. Precision was assessed by the coefficient of variation (CVsd), using the root mean square average. The limits of agreement were estimated from the differences between each replicate measurement of BMD and the estimated true value for a particular manufacturer, according to Bland and Altman. The results confirm the difference between devices from different manufacturers (18.5%). Mean CVsd values were 0.57% and 0.64% for Hologic and Lunar respectively. The limits of agreement among devices from the same manufacturer were 0.026 g/cm(2) and 0.025 g/cm(2) for Hologic and Lunar respectively. Differences in extreme results between devices from the same manufacturer were on average 5.4% and 3.6% for Hologic and Lunar respectively. Results of different devices from the same manufacturer are highly comparable, although unpredictable differences exist that may be clinically relevant.