Abstract : An elementary (major and trace elements) and isotopic (87Sr/86Sr, 143Nd/144Nd and δ18O) geochemical study has been achieved in order to determine and to quantify the petrogenetic mechanisms that gave rise to silicic magmas in Iceland. The target of this approach consists not only of understanding the genesis of these magmas but also in addressing their significance in terms of Iceland geodynamic evolution. The main results of this thesis are as follows: In course of Iceland history, most of the silicic rocks appear to have been generated by partial melting of the hydrated metabasaltic crust in a rift zone environment. In this place, the interaction between hot spot and mid-oceanic ridge is important resulting in a geothermal gradient high enough to exceed the solidus temperature of the partially hydrated basaltic crust. However, at the periphery of the island (i.e. Snæfellsnes Peninsula and Öræfajökull volcanic system), far from both the mantle plume centre and the rift-zone, the geothermal gradient is lower, precluding to step over the solidus temperature of hydrated metabasalts. This situation favoured the genesis of silicic magmas by fractional crystallisation (with possibly slight crustal assimilation).The link that exists between the composition of silicic rocks and the geodynamic environment of their genesis has been used together with age determination in order to establish the time-space geodynamic evolution of Iceland. The proposed model is based on the fact that mid-Atlantic ridge migrate relatively to the plume centre and on the "accretion - over lapping" mechanism that best accounts for the abnormally great width of Iceland. Iceland demonstrates that oceanic plateaus can give rise to significant volumes (towards 10%) of silicic magmas; in other words, it can be a potential environment for continental crust genesis in a purely oceanic environment. A comparison between these "continental" rock compositions from recent oceanic plateaus and primitive continental crust one (TTG) shows that, in spite of the Icelandic high geothermal gradients, which are assumed to be similar to the early Earth ones, modern oceanic plateaus cannot generate magmas having the primitive continental crust composition. Thus, it is concluded that oceanic plateau environment has not played a significant role, in the primitive continental crust genesis. Consequently, the "Iceland model" is not a modern analogue of the environment where proto-continental crust formed.