Two dual-energy X-ray absorptiometry (DEXA) systems--the Hologic QDR-1000 and the Norland XR-26 bone densitometers--were evaluated in terms of precision, accuracy, linearity of response, X-ray exposure, and correlation of in vivo spinal measurements. In vitro precision and accuracy studies were performed using the Hologic anthropomorphic spine phantom; linearity of response was determined with increasing thicknesses of aluminum slabs and concentrations of Tums E-X in a constant-level water bath. Both systems were comparable in precision, achieving coefficients of variation (CVs) of less than 1% in bone mineral content (BMC, g), bone area (cm2), and bone mineral density (BMD, g/cm2). Both were accurate in their determination of BMC, bone area, and BMD with reference to the Hologic spine phantom. Both systems also showed good BMC and BMD linearity of response. Measured X-ray skin surface exposures for the Hologic and the Norland systems were 3.11 and 3.02 mR, respectively. In vivo spinal measurements (n = 65) on the systems were highly correlated (BMC: r = 0.993, SEE = 1.770 g; area: r = 0.984, SEE = 1.713 cm2; BMD: r = 0.990, SEE = 0.028 g/cm2). In conclusion, both systems are comparable in terms of precision, accuracy, linearity of response, and exposure efficiency.