Long-term field experiments were conducted at six different sites (representing dryland and irrigated environments) with wheat (Triticum aestivum) on sandy loam soils to generate a database relating available water supply and fertilizer nitrogen (N) rate to grain yield. Step-wise multiple regression analysis showed that water supply and fertilizer N at more than 53 cm a−1 and 103 kg ha−1 a−1 were unproductive. A method, based on the principle of equi-marginal productivity of water in different periods of crop growth, has been used to estimate the optimum allocation of the variable amounts of water supply over the growing season. The optimized quantities of water in different periods of crop growth were then employed to compute fertilizer N requirement to achieve a given yield target. For medium grain yield targets (4–5 t ha−1), the range of substitution between fertilizer N and water for efficient resource use was quite wide.