The necessity to subdivide the wheat producing soils of the Winter Rainfall Region for extension purposes, was considered. Use was made of correlations, interaction variance components and pattern analysis procedures to reflect on the possible existence of sub-environments with different but consistent cultivar requirements. Yield data available at present seem to indicate primarily unpredictable sources of environmental variation to contribute to observed genotype×environment interactions. The implications for wheat extension programs are considered.S. Afr. J. Plant Soil1986, 3: 123–129