MESSENGER magnetometer data show that Mercury's magnetic field is not only exceptionally weak but also has a unique geometry. The internal field resembles an axial dipole that is offset to the North by 20% of the planetary radius. This implies that the axial quadrupol is particularly strong while the dipole tilt is likely below 0.8 degree. The close proximity to the sun in combination with the weak internal field results in a very small and highly dynamic Hermean magnetosphere. We review the current understanding of Mercury's internal and external magnetic field and discuss possible explanations. Classical convection driven core dynamos have a hard time to reproduce the observations. Strong quadrupol contributions can be promoted by different measures, but they always go along with a large dipole tilt and generally rather small scale fields. A stably stratified outer core region seems required to explain not only the particular geometry but also the weakness of the Hermean magnetic field. New interior models suggest that Mercury's core likely hosts an iron snow zone underneath the core-mantle boundary. The positive radial sulfur gradient likely to develop in such a zone would indeed promote stable stratification. However, even dynamo models that include the stable layer show Mercury-like magnetic fields only for a fraction of the total simulation time. Large scale variations in the core-mantle boundary heat flux promise to yield more persistent results but are not compatible with the current understanding of Mercury's lower mantle.

Various possibilities are currently under discussion to explain the observed weakness of the intrinsic magnetic field of planet Mercury. One of the possible dynamo scenarios is a dynamo with feedback from the magnetosphere. Due to its weak magnetic field Mercury exhibits a small magnetosphere whose subsolar magnetopause distance is only about 1.7 Hermean radii. We consider the magnetic field due to magnetopause currents in the dynamo region. Since the external field of magnetospheric origin is antiparallel to the dipole component of the dynamo field, a negative feedback results. For an alpha-omega-dynamo two stationary solutions of such a feedback dynamo emerge, one with a weak and the other with a strong magnetic field. The question, however, is how these solutions can be realized. To address this problem, we discuss various scenarios for a simple dynamo model and the conditions under which a steady weak magnetic field can be reached. We find that the feedback mechanism quenches the overall field to a low value of about 100 to 150 nT if the dynamo is not driven too strongly.