The purpose of the inverse electrocardiographic problem is to reconstruct the electrophysiological activities in the human heart from electrical signals measured on the body surface. This is a promising noninvasive approach to obtain an insight into cardiac diseases. However, this problem is illposed and regularization is required to stabilize the inverse solution. In the present work a spatio-temporal LSQR-Tikhonov hybrid regularization method is proposed, which combines the spatio-temporal and hybrid regularization frameworks. The novel method is tested in a realistic environment considering measurement noise, the modeling error induced by neglecting heart motion and baseline wander in ECG. The spatio-temporal hybrid regularization method achieves more stable results in the realistic environment compared to the Tikhonov regularization and the spatial LSQR-Tikhonov regularization. Moreover, the computation time is dramatically reduced thanks to the Greensite’s spatio-temporal approach.