Dual-photon absorptiometry (DPA) is a well-established procedure for measuring bone mineral density (BMD). Recently, dual-energy X-ray absorptiometry (DXA) has become available, which has the ability to measure BMD both regionally and in the total body (TB). We have evaluated the in vivo and in vitro precision of a DXA instrument and compared it with a DPA instrument with similar software characteristics. The short-term precision of BMD measurements using DXA was assessed in 65 postmenopausal women who had duplicate scans performed, with repositioning between scans. Precision was 0.9% in the lumbar spine and 1.4% in the femoral neck. The midterm precision of DXA was compared with DPA by scanning 10 volunteers a mean of four times over 24 weeks, on both instruments. The precision of the bone mineral content (BMC) and area measurements was significantly better (P less than 0.05) with DXA than with DPA. Long-term in vitro precision was assessed by scanning an aluminium spine phantom over 42 weeks, and a cadaveric sample over 52 weeks, on both instruments. Precision was similar using the aluminium phantom, but was significantly improved (P less than 0.001) when using DXA for scanning the cadaveric sample. Highly significant correlations (all P less than 0.001) of BMD, BMC and area measurements were observed when 70 volunteers were scanned on both instruments. However, there was a systematic difference in BMD values between the instruments. The precision of TB composition measurements assessed in 16 volunteers, over a 16-week period, were TB BMD 0.65%, TB lean tissue 1.47%, and TB fat tissue 2.73%.(ABSTRACT TRUNCATED AT 250 WORDS)