Maize is the staple crop for around 10 million people in the mid-hills of Nepal, where most families are in food deficit because maize yields are low and have remained stagnant for decades. The project achieved 30% yield increase over local varieties, in on-farm trials run and evaluated by farmers themselves and has initiated follow-on work to scale up the use of participatory approaches nationally. Although new varieties previously released have had higher yield potential than local cultivars, they have not been adopted by hill farmers because they were not compatible with the farming system. Maize is generally relay cropped with millet on crop terraces that are often affected by the presence of fodder trees. The trees reduce crop yield by altering the microclimate, affecting pest and disease incidence, and competing directly with the crop for light, water and nutrients. Farmers have sophisticated understanding of tree-crop interactions, and recognise local crop varieties that are tolerant of competition from trees and that are compatible with their integrated farming system. Farmers' knowledge about maize cultivation was systematically documented and then used to drive crop improvement, guiding modification and release of germplasm for testing by farmers and initiating change in agronomic recommendations. The knowledge base created during the project remains an enduring resource for agricultural improvement in the hills. Farmers' knowledge was verified in agronomic and physiological characterisation of the environment in which maize is grown, including the climate, soil and farmer decisions that define the context in which the maize crop has to perform. Local and introduced varieties were compared with respect to key traits that confer advantage in this environment, revealing on the one hand, deeper rooting of local over previously introduced varieties, indicating potential gains from combining traits through breeding, but no significant differences in photosynthetic light response. High yielding germplasm was obtained from CIMMYT, modified on-station in relation to farmers' preferences and then tested for local suitability by farmers themselves in participatory varietal selection (PVS) trials involving over a hundred farmers across five villages. This resulted in farmers adopting new varieties with greater yield potential, stress tolerance and other system compatibility benefits (such as stay green characteristics that improve the use of crop residues as animal fodder), improving average annual grain yields over traditional varieties by 28%, 16% and 32% in 1999, 2000 and 2001 respectively. The project harnessed the local knowledge of farmers, and their criteria for varietal selection, to incorporate system compatible traits from local maize with the higher yield potential of introduced germplasm in participatory plant breeding (PPB). Four PPB composites were created by crossing local and introduced germplasm. High yielding germplasm from yellow grained varieties was incorporated in a new white grained composite, important because white grain colour is a prerequisite for adoption. The best of these are now being tested by farmers in continued PVS and a project aimed at defining the requirements for scaling up the use of participatory crop improvement methods nationally through collaboration amongst government bodies and NGOs has been initiated.