To explore the genes differentiated between typical indica and japonica varieties, two typical indica/japonica varieties, Balilla (japonica) and Nantehao (NTH, indica), were selected to construct genetic populations based on the widely surveying for spikelet and pollen fertility of 90 indica/japonica F1 hybrids, which also were used as the wide compatability testers. In order to analyze the genes (QTLs) related to spikelet and pollen fertility, two reciprocal backcross populations Balilla/NTH//Balilla and Balilla/NTH//NTH were constructed and the spikelet and pollen fertility of each individuals were assessed. In both populations, two traits all appeared distorted normal distribution, but in the first population, they forwarded to low-level fertility type, the later population, forwarded to high-level fertility type relatively. The results indicated that both of male and female gametophytes of Balilla/NTH hybrids were partial sterile. Then we analyzed the SSR marker genotype of each individuals of Balilla/NTH//Balilla population containing 142 individuals, and constructed a SSR linkage map, in which, there were 108 information markers distributing on all 12 chromosomes equably, average marker distance was about 11.9 cM. Therefore the linkage map was qualified for QTL analysis. Two methods were employed to conduct QTLs analysis, i.e., single marker analysis and interval mapping. According to single marker analysis, 17 and 12 markers were found significantly responsible for spikelet and pollen fertility, respectively. And further study by means of MAPMAKER/QTL software, for spikelet fertility trait, two QTLs were detected, qSPTF1 on chromosome 1 and qSPTF6 on chromosome 6, and their additive effect were 13.501 and -16.414, respectively. According to previous studies, qSPTF6 was deduced to be the same locus as S-5. For pollen fertility, qPLLN7 on chromosome 7 and qPLLN9 on chromosome 9 were detected, and their additive effects were -12.003 and -11.012, respectively. Because the QTLs detected cannot explain completely the total variance of mapping population, other genetic factors must be existed to be responsible for spikelet and pollen partial sterility. Hence we let two random markers as putative covariates, and divide the 142 individuals into four groups according to the two marker genotypes, then the average values of spikelet and pollen fertility of each groups were calculated for two-way ANOVA (analysis of variance). The results indicated that there existed strong interaction for both spikelet fertility and pollen fertility. At a significance level of 0.005, there over 61 and 51 pairs loci interactions detected playing an important role in spikelet and pollen sterility expression, respectively. These results indicated that epistasis also was one of major genetic components controlling indica/japonica hybrid sterility.