An experiment was carried out for two consecutive years in order to evaluate the response of maize with the legume Canavalia ensiformis, and compared to a crop without legumes. The rotative plots were established at the beginning of the rainy season (May). In September of each year we planted anotherexperiment with rice. We evaluated two types of cultivation: one consisting of the incorporation of present legumes and weeds, present in the field before showing, and the other one consisted of leaving them on the surface. We evaluated four systems: one plot where we cut the legumes and left the waste on the site (CanR + F); another plot where the legume grew, and then we cutthe vegetative section ofthe plant and extracted the waste out ofthe plot (CanR): another plot with no legume and where we dumped the waste from the preceding plot (CanF), and the last plot with no legumes at all (SLeg). We found a highly significant response to the residual effect of the systems, and the group with the highest yielding was the CanR and also the CanR+F, with 5.39 and 5.28 tJha, respectively, followed by the Sleg plot, with a yield of2.50tJha. The response to N application was also very significant, wiht average yields of 2.82, 4.54 and 4.99 t/ha, for levels of 0,75 and 150 kg of N/ha, respectively. The interaction of N Doses X Sistems was highly significant.