AbstractAn effective cultivar development or testing program depends on the identification of nursery locations and testing sites that are representative of those encountered in commercial production. Defining unique production areas within a region often is complicated by the inability of the researcher to accurately determine the factors influencing crop response. Long‐term (1919–1983) county yield data for spring wheat (Triticum aestivum L.) were used as a basis for defining and characterizing production areas within North Dakota. Cluster analysis indicated that the state reasonably could be divided into five areas. These areas were characterized by differences in mean yield, standard deviation, range of yields, and stability of yield as measured by regression of county yield upon the state mean yield. A regression equation relating yield to distance north and east of the southwest corner of the state was used to construct yield isolines. The areas delineated by the cluster analysis and the pattern of the isolines were consistent with general physiographic areas of the state and climatic gradations. It was demonstrated that long‐term county yield data can provide a basis for defining production areas.