If coronary calcification scores obtained with electron-beam computed tomography (EBT) were proved to be correlated to coronary atherosclerosis, the reproducibility of the technique had to be assessed before being used for patient follow-up. A total of 150 patients, selected as a result of a cholesterol screening programme, were studied by EBT. Twelve contiguous 3-mm-thick transverse slices beginning on the proximal coronary arteries were obtained through the base of the heart. The amount of calcium was evaluated as the calcified area weighted by a coefficient depending on the density peak level. The value was expressed as a logarithmic scale. Intra-observer, inter-observer and inter-examination reproducibilities were calculated. They were 1.9, 1.3 and 7.2%, respectively. These results were good enough to allow the use of EBT for longitudinal studies. The influence of acquisition and calculation conditions on score computation were also analysed.