A dynamic bio-economic model is used to show that, without technological and policy intervention, soil nutrient balances, income and nutrition could not be substantially or sustainably improved in a highland area of Ethiopia. Although cash incomes could rise from a very low base by more than 50% over a twelve-year planning period, average per ha nutrient balances indicate significant nutrient mining and associated soil losses of about 31 tons per ha. With the adoption of an integrated package of new technologies (involving new high yielding crop varieties, agroforestry, animal manure and inorganic fertilizers use, construction of a communal drain to reduce water logging and some limited land user rights), results show the possibility of an average twoand- a-half-fold increase in cash incomes and a 28% decline in aggregate erosion levels over a twelve year period with a population growth rate of 2.3%. Moreover, a minimum daily calorie intake of 2000 per adult equivalent could be met from on-farm production, and per ha nutrient balances, while still negative for nitrogen and potassium, could be reduced by 36 and 6 % respectively, with phosphorous balances being reversed to positive values. However, these gains might be eroded by the need to meet increased nutritional demands arising both from increasing consumption levels and a more rapid population growth of over 2.8%. From a policy perspective, this reduction in nutrient losses in the face of higher reliance on the watershed for subsistence food requirements, would imply an increasing need for a more secure land tenure policy than currently prevailing, provision of credit to facilitate uptake of the new technology package and a shift from the current livestock management strategy that emphasizes use of livestock as a store of wealth to the one that encourages livestock keeping as a commercial activity. It would also imply a shift from a general approach to land management to a relatively more site-specific approach that recognizes the need for spatial and inter-temporal variability in input use based on land quality that would encompass an efficient nutrient management strategy.