Torsion of a long bone is the twist along its longitudinal axis; torsion of the radius is defined by the angle between the proximal and distal metaphyses in the transverse plane. Measurement of the radial torsion angle provides a means of detection and quantification of malrotation after a fracture. The purpose of the current study was to develop and standardize a technique for the measurement of torsion of the radius. Axial computerized tomographic images of thirty-nine pairs of dry cadaver specimens of normal radii, and an additional four pairs of radii with a unilateral deformity of the distal metaphysis that was consistent with a previous fracture, were studied and a measurement protocol was established. The radial torsion angle was measured by three independent observers on two separate occasions. Reproducibility of the technique was determined with use of the intraclass correlation coefficient to express both interobserver and intraobserver reliability. Consistency of measurements between observers and by the same observer was high, with intraclass correlation coefficients ranging from 0.87 to 0.94. The mean torsion angle for the eighty-two normal radii in the study was 32.6 degrees (95 per cent confidence interval of the mean, 30.3 to 34.9 degrees; range, 1.4 to 58.8 degrees). There were small variations in torsion angle between the two radii of each normal pair (mean side-to-side difference, 4.9 degrees; 95 per cent confidence interval of the mean, 3.5 to 6.3 degrees). The mean torsion angle of the four radii with a malunited fracture was 10.4 degrees (95 per cent confidence interval of the mean, 5.7 to 15.1 degrees), and the mean side-to-side difference in the pairs containing these radii was 24.1 degrees (95 per cent confidence interval of the mean, 8.5 to 39.6 degrees; p < 0.0001 compared with the normal radii).