The results of 196 clinical determinations of leg length inequality and postural pelvic tilt scoliosis in 21 patients were analysed and compared with reliable radiological measurements. Clinical methods proved to be inaccurate and highly imprecise, the observer error being +/- 8.6 mm for direct and +/- 7.5 mm for indirect measurement of leg length inequality, and +/- 6.4 degrees for the estimation of postural lumbar scoliosis. More than half (53%) of the observations were erroneous when the criterion of leg length inequality was 5 mm. Failure to determine the presence or absence of length inequality of more than 5 mm occurred in 54 measurements (27% of the total). In 12% of the direct and in 13% of the indirect measurements, the observers erred in deciding which leg was longer; discrepancies occurred even when radiological reading gave a leg length inequality of as much as 25 mm.