This thesis dealt with characterizing and modeling the dynamics of the three principal resources which are water, nitrogen and radiation in a no-tillage relay intercropping maize-cover crop systems, under sub-humid tropical area climate. Moreover, productivity of the main crop and of the whole system and nitrogen and water flows in the soil are assessed.The study was based on an on-field approach/experimental design carried out during two crop growing seasons in Cerrados region, in Unaí (Minas Géras, Brazil) since 2007. Firstly, an experimental station was used allowing an intense effort of measurement and secondly, a whole of several farmer's fields offering a wide range of soil and climate conditions and modalities of implementation of the studied cropping systems. In the latter, main crop was maize, and intercropped cover crop was a tropical fodder crop, either leguminous, Cajanus cajan or gramineous, Brachiaria ruziziensis. Two dates of sowing, early and late were considered for the sowing of the cover crop in established maize, in the case of the experimental station. This study showed that maize yield was not significantly affected by the presence of the relay cover crops in comparison with maize as the sole crop, even when the cover crop was sown soon after maize emergence in comparison with maize sole crop. In contrast, the production of biomass by the cover crop was significantly lower when grown with maize than when it was grown as a sole crop. In the intercropped systems, when sown early, the cover crop produced higher total biomass than when sown late (20 days after maize flowering). Moreover, total aboveground biomass production of maize intercropped with a cover crop was much higher than that of any of the crops sown alone. Total biomass produced by maize and pigeon pea (Cajanus cajan) was more than double that maize grown alone. The Land Equivalent Ratio (LER) values of both maize grain yield and biomass production were higher than one, whatever the intercrop system, reaching up to 2.03 providing better available resources use efficiency by the intercropped plants. Thus, such systems permit to produce both maize grains and forage.The thesis shows also the strong variability of the production potential of these systems in farmer's fields and suggests that when water and nitrogen resources are less available than in the experimental station, competitions between maize and cover crops can reduce the agronomic performances of maize and/or the cover crop when compared to the equivalent pure sole crop. Total biomass produced by intercropping remained however higher than that of maize sole crop, except for rare exceptions where lower resources availability, sometimes accentuated by difficulties of technical practice, maize seems to suffer from stronger competition and/or the cover crop produced only very little biomass. The thesis evaluates the capacity of the STICS intercrop model, STICS-CA, to provide a finer analysis of the dynamics of the resource sharing in these systems and to estimate their agronomic performances and certain environmental impacts according to technical practices and soil and climate conditions that were different from those of our experimental station. Minor modifications and model calibration led to obtain satisfying simulations of the sole crops maize and Brachiaria and intercrop maize-Brachiaria in no-tillage under mulch. On the other hand, it was not possible to obtain satisfying simulations of sole crop maize in conventional tillage, nor of sole crop pigeon pea and intercrop maize-pigeon pea in no-tillage under mulch. The thesis provides information on the model imperfections and as well proposes improvements in terms of formalisms to take into account the mineralization of the mulch, as in terms of experimental design.