Precipitates of calcium phosphate in coronary arteries always indicate the presence of atherosclerosis. The mass of these precipitates is related to the severity of atherosclerosis. To determine the accuracy of electron beam computed tomographic (CT) mass estimates, we imaged 21 human hearts inside an anthropomorphic chest phantom using an Imatron C-100 electron beam CT scanner (Imatron, San Francisco, CA). We then incinerated the 63 imaged coronary arteries and weighed the mineral ash. We calculated the mass estimates from the images using an algorithm derived from a model that assumes simple radiographic properties of the coronary arteries. We also calculated the currently used coronary calcium score for each artery. Although both the mass estimates and the scores correlated with the actual mass of the incinerated specimens (r = .97 and r = .93), the correlation with the mass estimates was better (P = .02; William's test). The regression equation relating the actual mass to the mass estimates was y = 1.37 x + 14, indicating that the CT mass estimates consistently underestimate actual coronary calcium phosphate mass. We conclude that relative mass estimates using electron beam CT scanning are accurate and that both these and the currently employed calcium scores reflect the actual mass of precipitated calcium phosphate in diseased coronary arteries.