Magnetocardiography (MCG) is a non-invasive diagnostic tool used to investigate the activity of the heart. For applications in an unshielded environment, in order to extract the very weak signal of interest from the much higher background noise, dedicated hardware configuration and sophisticated signal processing techniques have been developed during the last decades. Being powerful in noise rejection, the signal processing may introduce signal distortions, if not properly designed and applied. However, there is a lack of an effective tool to quantitatively evaluate the signal integrity for MCG at present. In this paper, we have introduced a very simple method by using a small coil driven by a human ECG signal to generate a simulated MCG signal. Three key performance indexes were proposed, which are correlation in time domain, relative heights of different peaks and correlation in frequency domain, to evaluate the MCG system performance quantitatively. This evaluation method was applied to a synthetic gradiometer consisting of a second-order axial gradiometer and three orthogonal reference magnetometers. The evaluation turned out to be very effective in optimizing the parameters for signal processing. In addition, the method can serve as a useful tool for hardware improvement.