To understand the genetic basis of seed setting rate (SSR) in rice, two breeding lines (Oryza sativa ssp. indica), T226 with higher and stable SSR and T219 with lower and fluctuating SSR from different natural conditions, were used for constructing recombinant inbred lines (RILs). Genotype by environment (G×E) interaction and quantitative trait loci (QTL) for SSR were analysed using a population with 202 RILs under eight differing environments. A significant G×E interaction for SSR was detected in rice using the Additive Main Effects and Multiplicative Interaction (AMMI) statistical model, and the IPCA1 and IPCA2 of the G×E interaction accounted for a variation of 57.6%. QTL controlling the SSR were detected by the method of interval analysis. Seventeen QTL on nine chromosomes were identified across eight environments, totally explaining the phenotypic variances from 4.6 to 35.7%. Most of the QTL, each explaining a small part of the phenotypic variances and interacting with environments, were detected in one or two environments, and their alleles for increasing the SSR were derived from T226. However, the QTL (MRG5959–MRG2180) on chromosome 3 was detected across six different environments. It explained maximum phenotypic variances in each detected environment and its allele for increasing the SSR was derived from T226. Another QTL, mapped between markers RM592 and RM169 on chromosome 5, was detected in five various environments and its allele increasing the SSR was derived from T219.