Magnetocardiographic and body surface potential mapping data measured in 6 patients with multivessel coronary artery disease were used in equivalent current-density estimation (CDE). Patient-specific boundary-element torso models were acquired from magnetic resonance images. Positron emission tomography data registrated with anatomical magnetic resonance imaging data provided the gold standard. Discrete current-density estimation values were computed on the epicardial surface of the left ventricle from difference (stress-rest) ST-segment maps. The ill-posed inverse problem was regularized with 3 different methods (Tikhonov regularization with an identity or a surface Laplacian operator and a maximum a posteriori estimator). Comparisons with positron emission tomography studies showed that the maximum a posteriori estimator is superior to other regularizations, provided that a suitable a priori information is available. In general, good correspondence was found for segments of high and low amplitude in current-density estimations, and the viable and scar areas in positron emission tomography, respectively.