The accuracy of radiographic measurements of lumbosacral lordosis, vertebral size, spondylolisthesis and disc height were studied. Errors caused by lateral tilt or longitudinal axis rotation of the lumbosacral spine were assessed in an experimental study on 12 cadavers. In a series of 170 normal individuals, the distribution of radiographic variables describing lumbosacral lordosis, size of vertebrae and disc height was investigated. The interobserver measurement variation was studied with respect to all measured variables. The differences between radiographs of 125 spondylolytic patients in the recumbent and standing positions were analyzed with respect to vertebral slipping and lumbosacral lordosis. There were only minor projectional and interobserver measurement errors in the variables describing vertebral size and lumbosacral lordosis, which make these variables suitable for radiographic assessment at repeated examination. The absolute measurements of disc height were too inaccurate to be used as a basis for conclusions. A semiquantitative approach is more reliable, and can be recommended for the comparison of radiographic assessments of disc height. In adults, the radiographic evaluation of vertebral slipping and lumbosacral lordosis is equally reliable in the recumbent and standing positions.