Recordings of the magnetic fields (MFs) arising from gastric electrical activity (GEA) have been shown to be able to distinguish between normal and certain abnormal GEA. Mathematical models provide a powerful tool for revealing the relationship between the underlying GEA and the resultant magnetogastrograms (MGGs). However, it remains uncertain the relative contributions that different volume conductor and dipole source models have on the resultant MFs. In this study, four volume conductor models (free space, sphere, half space and an anatomically realistic torso) and two dipole source configurations (containing 320 moving dipole sources and a single equivalent moving dipole source) were used to simulate the external MFs. The effects of different volume conductor models and dipole source configurations on the MF simulations were examined. The half space model provided the best approximation of the MFs produced by the torso model in the direction normal to the coronal plane. This was despite the fact that the half space model does not produce secondary sources, which have been shown to contribute up to 50% of the total MFs when an anatomically realistic torso model was used. We conclude that a realistic representation of the volume conductor and a detailed dipole source model are likely to be necessary when using a model-based approach for interpreting MGGs.