One of the principal research axes related to nuclear energy concerns the management of nuclear wastes. In this context, certain innovative pathways are being explored to provide options which could lead to a major reduction of the quantities of radioactive nuclides requiring long-term geological storage. An important aspect is the development of advanced nuclear systems which allow the transmutation of long-lived minor actinides, the ADS (accelerator driven system) concept often being proposed for the purpose. Various important developments are necessary for these new systems, one of the principal needs being the qualification of calculational tools to enable the reliable design of an ADS demonstrator. In the present research, related numerical and experimental studies have been carried out in the framework of the MUSE4 programme, which was developed on the MASURCA facility at Cadarache, France. This programme consists of investigations of a reference critical configuration (MOX fuel with sodium), and several subcritical configurations driven by an external neutron source. The source is obtained with the help of a deuton accelerator, called GENEPI, which produces neutrons via a nuclear reaction, either with a tritium target (14.1MeV neutrons), or with a deuterium target (2.7MeV neutrons). The experimental techniques implemented in this research address the study of spectral variations via foil activation measurements near interfaces characteristic of ADSs. The additional value of such experimental results, compared to standard fission rate traverses obtained using fission chambers (which were also analysed), has been highlighted. An important novel set of integral parameters has thus been made available for the validation of different numerical codes and nuclear data. The activation and fission rate measurements carried out, as also experimental results for other parameters such askeff, βeff and φ*, have been interpreted using various calculational codes (mainly ERANOS and MCNP) and several different nuclear data libraries (JEF-2.2, ERALIB1, ENDF/B6v2). Supplementary information on the performance of the calculational methods has been obtained in the context of the MUSE4 benchmark exercise, conducted by the NEA. The present investigations have led to the resolution of various problems encountered during the earlier MUSE3 programme at MASURCA, viz. uncertainties related to description of the intrinsic source description, difficulties in interpreting fission traverses near the external source, etc. A detailed study has thus been made possible of the important factors which characterise subritical fast systems driven by an external source, i.e. influence of the subcriticality level and of the external source intensity. The various analyses carried out have permitted to establish several recommendations concerning the modeling of the external source and the accelerator tube, as well as the most appropriate numerical methods and data sets to be applied. In this context, the specificity of subcritical configurations, as regards the prediction of parameters such as spatial indices and absolute reaction rates, has been clearly brought out. Certain difficulties (interpretation of high threshold reactions, absolute flux determination, etc.) have thereby been highlighted, pointing at the need for further numerical and experimental investigations. Finally, some recent ERANOS developments have allowed a study to be carried out of the representativity of MUSE4 configurations with respect to alternative ADS concepts proposed in the framework of the international PDS-XADS project. In brief, the representativity of the MUSE4 configurations is found to be quite satisfactory for XADS-Na and XADS-gas. However, in the case of the XADS-Pb/Bi system, effects of the significant uncertainties associated with the nuclear data of lead and bismuth have been clearly highlighted, indicating the desirability of new, more specific experiments for XADS-Pb/Bi validation.