Lumbar spine bone mineral density (BMD) was measured by quantitative digital radiography, a new dual energy x-ray technique, and by 153Gd dual photon absorptiometry (DPA) in 85 patients. Each patient was measured twice by the new method and once by DPA on the same day, with repositioning between measurements. Serial measurements were made on an hydroxyapatite spine phantom embedded in tissue-equivalent plastic to evaluate the long term reproducibility of each instrument. The spinal BMD measurements with the 2 techniques were linearly related and highly correlated (r = 0.98) over a range from severely osteopenic to high normal. This correlation was not affected by the age, weight, or BMD of the patient measured. Quantitative digital radiography's long-term reproducibility using the spine phantom was stable for 180 days (coefficient of variation, 0.23%); DPA values were 3 times as variable for 170 days (coefficient of variation 0.73%) and increased 1.0% (P less than 0.0001) after a software change. The short term reproducibility of quantitative digital radiography, estimated from paired patient measurements, was 2-fold better than reported values for DPA and was independent of the patient's age, weight, or BMD. Measurement time by quantitative digital radiography was 5-8 min, with a maximum radiation exposure of 3 mrem, significantly lower than the corresponding DPA values. Quantitative digital radiography's image resolution was superior to that of DPA, enabling it to measure more bones. These advantages along with the elimination of 153Gd source changes and Nuclear Regulatory Commission licensing requirements indicate that quantitative digital radiography is the superior method for spinal BMD measurements.