Few reports on the distribution of resistance genes to plant pathogens within a species are available. Because of its genetic structuration, Oryza sativa can be an interesting model for such a study. We were interested in the interaction between the resistance gene to blast Pi33 and its corresponding avirulence gene ACE1. Although Pi33 was not cloned, its allele(s) conferring resistance can be identified by inoculation of pairs of isogenic strains of Magnaporthe grisea differing only for ACE1. One-hundred-eighty-three varieties of O. sativa were inoculated with isogenic strains. Twenty-three varieties of the Indica subspecies (mainly modern semi-dwarf varieties) carrying a resistance allele of Pi33 were identified representing 21,6 percent of the Indica varieties tested. None of the 45 Japonica varieties tested carried the resistance allele. In order to identify the original donors of Pi33, the genealogies of the varieties carrying Pi33 were examined. In addition, 28 parental lines from IR64 were tested to seek the donor of resistance for this variety. Two possible origins of resistance could be identified: Tsai Yuan Chung and the wild rice O. rufipogon. These two candidates were confirmed as possible sources of resistance by genotyping and sequencing. It appears that both resistance sources were commonly used in breeding programmes leading to modern semi-dwarf Indica varieties. The presence of Pi33 in O. rufipogon led us to the hypothesis that Pi33 existed before domestication. This hypothesis was confirmed by the detection of Pi33 in O. latifolia (CCDD genome), O. barthii (AA) and diverse accessions of O. rufipogon (AA).