The release of mineral-N in soil from plant residues is regulated by their ‘quality’ or chemical composition. Legume materials used by farmers in southern Africa are often in the form of litter with N concentration <2%. We investigated the decomposition of Sesbania sesban and Acacia angustissima litter in the field using litterbags, and N mineralization of a range of legume materials using a leaching tube incubation method in the laboratory. The mass loss of the litter could be described using a modified exponential decay model: Y = (Y0−Q)e−kt + Q. The relative decomposition constants for Sesbania and Acacia litter were 0.053 and 0.039 d−1, respectively. The % N mineralized from fresh Sesbania prunings was 55% compared with only 27% for the Sesbania litter after 120 days of incubation under leaching conditions. During the same period, fresh prunings of Acacia released only 12% of the added N while Acacia litter released 9%. Despite the large differences in N concentration between Acacia prunings and its litter, the total mineralized N was similar, as mineralization from prunings was depressed by the highly active polyphenols. While N supply may be poor, these slow decomposing litter materials are potentially useful for maintaining soil organic matter in smallholder farms. In two field experiments with contrasting soil texture, Sesbania, Acacia and Cajanus produced large amounts of biomass (>5 Mg ha−1) and improved N cycling significantly (>150 kg N ha−1) on the clay loam soil, but adapted poorly on the sandier soil. There was a rapid N accumulation in the topsoil at the beginning of the rains in plots where large amounts of Sesbania or Acacia biomass had been incorporated. Despite the wide differences in resource quality between these two, there was virtually no difference in N availability in the field as this was, among other factors, confounded by the quantity of N added. A substantial amount of the nitrate was leached to greater than 0.4 m depth within a three-week period. Also, the incidence of pests in the first season, and drought in the second season resulted in poor nitrogen use efficiency. Our measurements of gaseous N losses in the field confirmed that N2O emissions were <0.5 kg N ha−1. As we had measurements of all major N flows, we were able to construct overall N budgets for the improved fallow – maize rotation systems. These budgets indicated that, in a normal rainfall season with no major pest problems, reducing nitrate leaching would be the single largest challenge to increased N recovery of added organic N in the light textured soils.