This paper summarized the theory of welfare economics and natural resource allocation, and illustrated how application of this theory can lead to Pareto improvements in the design of irrigation systems in tropical arid- and semi-arid agricultural systems. Drawing on the two examples of nitrate losses and salinization, the paper concludes that economic analysis can help design policies to conserve natural resource stocks, but that such analysis is limited by insufficient data, methodological difficulties, and by a lack of integration between natural and social research. A portion of the paper discusses the physical and biological processes associated with nitrate losses and salinization, and review research on their magnitudes in arid- and semi-arid zones. This review demonstrates that certain combinations of hydrological conditions and farming systems will likely result in nitrate losses and/or salinization. The design of irrigated systems in these cases might be improved by analyzing the externalities associated with these phenomena. There are two policy approaches - curative and preventive - to mitigate losses from nitrate pollution and salinization. Preventive policies, which entail directive measures (e.g., standards) or the creation of economic incentives and disincentives, are generally more cost-effective and risk-reducing. Taxes, however, should be avoided where natural resource demand is inelastic. Combining subsidies with taxes allows greatest flexibility in cost distribution, but reduces the net benefits from mitigation. Given the institutional weakness of most lower income countries, governments attempting improved management of natural resource stocks must rely primarily on conservation subsidies rather than on standards, taxes, or curative measures. More generally, informed judgments on the magnitude of externalities associated with any one natural and social environment depend on information being available to economists. Yet comparative data on agro-ecological research should be designed to illuminated the policy implications of natural resource management. The effects of irrigation on soil fertility and on water quality are two high-priority topics. Economists need to evaluate the contributions of natural resource quality to individual well-being. Policy makers meanwhile may choose to establish qualitative standards for important natural resources, and to distinguish among broad categories of natural resources based on their substitutability. Economists also must understand how current policies generate farm-level incentives for overuse of natural resources, since corrective policies may vary for different incentives. Finally, economists should closely examine the degree to which the use of high discount rates in economic analysis leads to environmental degradation by reducing the value of future benefits and costs.