The inverse ECG problem is ill-conditioned and its solution requires a relatively high computing effort. Additional constraints are required in order to obtain a stable solution. A method is proposed in which the solution of the inverse ECG problem is approached in the frequency domain, taking advantage of the assumption that propagation delays may be ignored and the quasi-periodicity of ECG. In this method usual Tikhonov zero-order constraints are applied to the amplitudes of the signals for a selected frequency domain. This method ensures faster solutions that are spatially and temporally well behaved. Calculation of epicardial electrograms is compared to a basic method.