The results of 48 years of oat stem rust physiologic race surveys are interpreted in terms of the prevailing host genotype. Changes in the frequencies of genes that govern virulence in the pathogen can be explained only in part by changes in the resistance genes carried by the host population. Genes for virulence on newly released types of resistance have spread very quickly through the rust population, after initial 'breakdown' of the resistance. The most successful physiologic races carry genes for virulence in excess of those required for successful parasitism in North America. Many races carry genes for virulence on types of resistance that have never been used on this continent.