In this thesis work I have been concerned with modeling the magnetic eld of planet Mercury. With this goal in mind, I tested a new method for modeling the time dependent magnetic eld of a planet, the Time Variable Equivalent Source Dipole method. I showed that in the case that applies to Mercury, namely a concentration of observations over the northern hemisphere, this method gives better results than the standard Spherical Harmonic method. The main conclusion of this thesis is that, assuming that the secular variation of Mercury is around one tenth of its main eld, in amplitude, the new developed method will allow to recover it from MESSENGER data. This has very important consequences for the understanding of the core dynamics and in particular of the dynamo mechanism that generates Mercury magnetic eld. This work is nished, but the problem is not solved. Nevertheless, it clari ed di erent questions that will make it easier to continue for the analysis of the true measures of MESSENGER spacecraft.