The accuracy of measurement of the bone mineral content (BMC, g) and bone mineral density (BMD, g/cm2) of the lumbar spine by dual photon absorptiometry (DPA) was estimated by means of two different spine scanners (a Nuclear Data 2100 and a Lunar Radiation DP3). The lumbar spines of 13 cadavers were used. BMC and BMD were measured in situ and on the excised vertebrae in a solution of water/ethanol; and covered with ox muscle/porcine muscle/lard. The actual mineral weight and areal density were determined after chemical maceration, fat extraction, drying to a constant weight, ashing for 24 hr at 600 degrees C, and correction for the transverse processes. The true are was measured by parallax free X rays and planimetry. All measurements of BMC or BMD were highly interrelated (r = 0.94-0.99). The standard error of estimate (s.e.e.) of BMC in situ versus BMC in water/ethanol was 5.2%. The agreement between the BMD values of the two scanners was very good (s.e.e. = 2.9%). BMC in situ predicted the actual vertebral mineral mass with an s.e.e. of 8.1%. BMD in situ and BMD in water/ethanol predicted the actual area density with s.e.e.s of 10.3% and 5.0%, respectively. This study discloses the correlation and accuracy error of spinal DPA measurements in situ in whole cadavers versus the actual BMC and BMD. The error, which is underestimated in in vitro studies, amounts to 10%.