Stripe (or yellow) rust caused by Puccinia striiformis f. sp. tritici is the most destructive foliar disease of wheat in China. The pathogen populations were analyzed for virulence evolution, complexity, phenotypic dynamics, and diversity on temporal and spatial bases. A total of 41 races were identified and characterized from 4,714 stripe rust isolates collected during 2003 through 2007 from wheat growing areas in 15 provinces in China. The races were based on avirulence/virulence patterns to 19 differential host genotypes. Chinese stripe rust population exhibited high diversity with a complex virulence structure. Comparisons using the relative Shannon's index indicated that some differences in the richness and evenness of races were present in pathogen populations within years and between regions despite a national tendency to reduced diversity over time. A noticeably increased frequency of race CYR33 (Chinese yellow rust 33) with virulence for YrSu was the major virulence change recorded in this study compared to the results on an annual basis. Isolates of Puccinia striiformis f. sp. tritici from different regions showed differences in the composition of races, distribution frequency, and diversity. The uneven distribution of major races and comparatively greater diversity in the Northwest and Southwest regions than that in the Huang-Huai-Hai region suggest that long-distance migrations of the pathogen occur from one or more over-summering areas eastward into over-wintering areas. This supports the hypothesis that southern Gansu and northwestern Sichuan comprises a "center of origin for virulence". Mutation of virulence or avirulence for host resistance in the stripe rust fungus may be the basic cause of the occurrence of new virulent types. The subsequent dominance of certain races will vary with parasitic fitness and the opportunities to be selected through large-scale cultivation of varieties with matching resistance genes. Implications of the center of origin for virulence variation and diversity in the pathogen population and an alternative strategy for limiting virulence evolution are discussed.