Abstract : This thesis work consists of two parts: a) theoretical, and b) experimental. We combine and use the high energy transport code LAHET, the low energy transport code MCNP, and the activation code CINDER. Our benchmarking calculations show that LAHET neglects the Coulomb dissociation for deuterons. By adding this missing term, we obtain a good agreement with the available data. We also conclude that LAHET describes well the data for isotope production yields if the ORNL fission model is employed for nuclei with Z>90. The 'default' RAL fission model gives too broad isotopic distributions and fails to reproduce the data in absolute value. We examine different combinations of beams, beam energies, spallation target and multiplying medium materials in order to optimize the neutron production, energy amplification and isotope production via neutron induced fissions. We show that the (d,xn) reactions could bring a number of important advantages when compared to the (p,xn) reactions. We conclude that the use of deuterons instead of protons should result in higher primary beam intensities, lower costs of the system and facilitate radioprotection problems. Within the SPIRAL Phase-II project at GANIL, we propose d(100 MeV)+Be→xn+U as an optimum combination for the production of neutron rich nuclei in the mass region 75≤A≤160. However, the production of tritium gas in the target-converter should be considered carefully. The use of heavier metal targets-converters may cause more severe problems of radioprotection.Our experimental work is closely related to the theoretical investigations. We measure the complete proton spectra for 1.00 and 200 MeV deuteron induced reactions on 8 thin targets (Be, C, Al, Ni, Nb, Ta,, Pb and U) and in the angular region 8 deg C ≤ θp ≤ 120 deg C. The experiments were carried at LNS (Saclay, France) and at NAC (Faure, South Africa). Good quality data (within 10% in absolute value and with 4-8 MeV energy threshold) support our improved LAHET physics modelling for (d,xp) and, consequently, for (d,xn) reactions.