Abstract: Indica-japonica hybridization is one of the most important breeding methods in China, whereas identifying subspecies differentiation mechanisms is the key in indica-japonica hybridization breeding. By using InDels (Insert/Deletion) and ILPs (Intron Length Polymorphism), an analysis was made on the F₆ populations derived from the hybridization of indica-japonica (Qishanzhan/Akihikari) planted in Liaoning and Guangdong provinces and generated by bulk harvesting (BM), singleseed descent methods (SSD), and pedigree method (PM). No segregation distortion was observed for the BM and SSD populations. The frequency distribution of japonica kinship percentage (Dj) was concentrated in 40%-60%. The PM populations in the two provinces presented indicadeviated distribution (30%-55%), with significant difference between Guangdong (38%) and Liaoning (42%). In addition, there was a significant positive correlation between the Dj and the kinship of functional gene regions in the BM and SSD populations. However, part of the positive correlation was broken in the PM populations that showed a regular distribution in the genotype patterns of indica and japonica loci. The above results demonstrated that artificial selection could be the main factor affecting the population differentiation in indica-japonica hybridization, and, with the synergistic effect of natural selection, induced the phenomenon of segregation distortion. There existed a close relationship between the differentiation of subspecies and the important agronomic traits, which could be the main reason why indica-japonica hybridiation breeding could not achieve the expected effect of combining the two subspecies advantages.