Abstract : Nearly 60 % of the Earth surface is composed by basaltic rocks which form the upper part of the oceanic crust. These PhD works focus on the elastic properties of basalt at several spatial scales : field scale, sample scale and porosity microstructure scale. With its active hydrothermalism Iceland appears as the ideal natural laboratory to study the interactions existing between basalts, fluids and fracturing at a large scale. We first begin by the study of a paleosite of hydrothermal fluid flow which took place in a basaltic pile. Some evidence of hydrofracturing are identified. We also investigate the porosity of the different materials associated to the hot fluid flow. The second part of the field study is based on microseismicity surveys on the Reykjanes Peninsula which is an active area in Iceland. The most important results is those deduced from inverting the tomography data obtained by C. Dorbath in 2005. By applying an effective medium theory to seismic velocities we have attempted to estimate the crack density and aspect ratio of the Icelandic crust at this place. We have show that areas characterized by a strong P-waves velocities anomaly were characterized by high crack density and low aspect ratio at 6 km depth. We have also investigated the effect of the fluid compressibility on the crack parameters with depth. The sample scale is investigated through two studies. The first one is the investigation of three different modes of deformation in an Icelandic basalt by using laboratory seismological tools (elastic waves and acoustic emissions) and scan imaging. first of all we show that at low effective pressure (5 MPa) an axial loading induces a shear failure in the basalt with a classical angle of about 45°. On the contrary at higher effective pressures (75 MPa and more) an increasing of the axial stress induces a localization of the deformation in the centre part of the sample. Focal mechanisms of the acoustic emissions reveal an important part of compression events (mode I rupture) suggesting pore collapse mechanisms. Such compaction structures are usually obtained for porous rocks (20-25 %). And yet the investigated basalt has an intact total porosity of about 10 %. Then the size of the equant porosity and its ratio with grain size can be involved for explaining the pore collapses. Finally the third triaxial experiment is an induced fluid pressure failure from a high confining pressure state (80 MPa). A large shear plane failure is formed due to pore pressure increasing the local porosity of about 1 %. The second sample scale study is based on the porosity scale in order to investigate the frequency effect on elastic moduli. To obtain experimental data we performed hydrostatic experiments on an Icelandic basalt specimen under both dry and saturated conditions. This basalt is characterized by a bimodal porosity, i.e., cracks and equant pores. The elastic properties -bulk moduli in our case- were investigated under high pressure through two experimental methods : (1) a classical one using ultrasonic P- and S-waves velocities (frequency 106 Hz), (2) and a new one, using oscillation tests (frequency 102 Hz). In dry condition, experimental data show no significant difference between high (HF) and low (LF) frequency bulk moduli. However, in saturated conditions, two effects are highlighted : a physico-chemical effect emphasized by a difference between drained and dry moduli, and a squirt-flow effect evidenced by a difference between HF and LF undrained moduli. The experimental approach was completed by a theoretical study. The HF moduli are derived from a new effective medium model with an isotropic distribution of pores and cracks with idealized geometry, respectively spheres and ellipsoids. LF moduli are obtained by taking HF dry moduli from the model and substituting into Gassmann's equations. In the case of a porosity only supported by equant pores, the calculated dispersion in elastic moduli is equal to zero. In the case of a crack porosity, no bulk dispersion is predicted but a shear dispersion appears. Finally in the general case of a mixed porosity (pores and cracks), dispersion in bulk and in shear is predicted. Our results show that the maximum dispersion is predicted for a mixture of spheroidal pores and cracks with a very small aspect ratio (<10-3). Our theoretical predictions are compared to experimental data and a good agreement is observed. We also used our theoretical model to predict elastic waves velocities and Vp/Vs ratio dispersion. We show that the P-waves dispersion can reach almost 20 % and the Vp/Vs dispersion a maximum value of 9 % for a crack porosity of about 2 %. Since laboratory data are ultrasonic measurements and field data are obtained at much lower frequencies, these results are useful for geophysicists to interpret seismic data in terms of fluid and rock interactions.