This research employed a wide range of approaches to address the potential for improving soil fertility management and productivity in pearl millet systems in northern Namibia. Emphasis was placed on improving the management of grain legumes and their contributions to soil fertility and system productivity and so directly addressed the major goal of DFID: poverty elimination. Research activities comprised a programme of on station and on farm field research supported by a literature review, a number of different survey and mapping exercises and nutrient budget modelling. New grain legume germplasm was accessed from a number of sources outside Namibia and introduced into the existing Namibian legume screening programme. The project adopted a predominantly on-farm form of new varietal screening that promised to identify varieties more robustly suited to on farm conditions and farmers' needs than is possible with on station screening. The research indicated that cowpea had particular potential for making greater contributions to soil fertility but that the most likely route to legume assisted productivity improvements is through the adoption of a number of different legume management options that cumulatively have an impact on soil fertility. The modelling activities suggested that the adoption of a single management practice involving a legume is unlikely to make a substantial contribution to soil fertility. The modelling also revealed the importance of (legume) residue management and nitrogen fixation rates in determining whether the legume was a net nitrogen contributor or miner of the system. 5 Promising new legume technologies identified included: - cowpea/millet intercropping with increased cowpea densities, - increasing the productivity of millet/cowpea intercrops with modest phosphorus and manure additions - the use of cowpea and other short duration grain legumes as green manures for pearl millet There is great potential for farmers to work in adapting legume species and varieties to fit the complex biophysical and management niches that exist in their farming systems. From the project experiences it is clear that more needs to be done in facilitating the access of farmers to new legume germplasm and some recommendations are made for changes in the approach to grain legume screening in Namibia. The project collaborated on work on systematising indigenous systems for land classification and combined them with more formal systems based soil characteristics. This has yielded a very useful GIS map-based framework for understanding the complexity of the biophysical environment in the study areas. This promises to be a very useful tool in research planning and targeting research activities and advice to individual fields, farms and households.