This paper briefly reviews examples of the application of crop simulation models to develop more practical fertility management options in Zimbabwe and Kenya. The work in Kenya also considers alternative strategies for managing water runoff. The paper reviews why and how simulation models are being employed, and outlines some of the problems arising during the early stages of application. The practical targets for crop simulation modeling are reviewed. The semi-arid cropping systems of southern and eastern Africa are characterized by their low productivity. Cereal grain yields average less than 750 kg ha-1 in most countries in the region. Despite the increasing adoption of new varieties of maize, sorghum and pearl millet, productivity levels remain largely unchanged. Most households commonly fail to produce enough grain to meet their requirements. In drought years, most farm families depend on food aid and supplementary grain purchases (Hedden- Dunkhorst 1993). A major reason for the continuing low productivity of the semi-arid cropping system is the failure of farmers to adopt improved fertility management practices. Less than 5% of these small-scale farmers use fertilizer. Most have never even tried this input. A surprising proportion of farmers (e.g., 50% in southern Zimbabwe) fail to use available manure. In consequence, the quality of soil resources is declining. The limited adoption of chemical fertilizer is surprising in view of the consistency of trial results indicating the potential profitability of this input (e.g. DR&SS 1992; DR&SS 1993; Gono 1996). One explanation is that fertilizer is not readily available in rural, retail shops. Farmers must travel long distances to find this input and then incur high transport costs to carry this back to the farm. Adoption is also constrained by the difficulty farmers have in assessing the value of fertilizer. Use of this chemical input is undoubtedly risky in the semi-arid environment. However, growing evidence suggests that farmers have over-estimated these risks. Indeed, the variability of rainfall characteristic of these environments makes it difficult to accurately assess the level and likely range of investment returns. These difficulties are compounded by the tendency of national research and extension services to recommend rates of fertilizer use well above the levels farmers are willing to consider for their own experimentation. Many countries still offer blanket national recommendations without regard for the variability of their rainfall or soil conditions. Yet even when recommendations are adjusted for the semi- arid agroecology, the prescribed levels of application are greater than the levels of investment farmers are willing to consider. In effect, these recommendations are irrelevant to most farmers. But they also communicate to most farmers that fertilizer is an impractical input.