Dispersed, simple fertilizer trials on irrigated wheat crop were laid out in farmers' fields spread over 280 locations in Afghanistan extending over three seasons during 1970 to 1973. The 219 locations from which data could be collected were partitioned statistically into three arbitrary soil fertility classes by juxtaposing the “compound Baule” computed from soil test data with the corresponding check plot yield in each location. Adopting a graphical procedure, we fitted a three-nutrient Mitscherlich-Baule generalized yield equation to the grain yield data; we also worked out the economic optima in respect of “maximum profit” and “maximum return” considerations and the corresponding fertilizer rates for each group of soils. We found that the potential maximum yield of grain wheat on all types of soils in the country was estimated as 6232 ± 827 kg per ha, and that the greatest economic advantage would accrue if most of the available fertilizers could be applied to the less fertile soils. Regression equations have been presented for estimating the β parameter of the generalized yield equation from soil test values, in order that fertilizer rate recommendations could be formulated for other, similar areas in the country and for the situations when the cost/price ratio changes or management norms improve.