Rice, which has a modest genome size, is a wonderful model crop that has been comprehensively studied in functional genomics. With vast germplasm resources and reduced sequencing cost, genome-wide association study (GWAS) has been feasible to explore the allelic variation underlying agronomic traits in rice. In this review, we firstly summarize the work on constructing GWAS platforms in human genetics and in Arabidopsis and maize, from which we have a lot to learn. We then discuss the sequencing strategies and experimental designs for GWAS of agronomic traits in rice, and the perspective and challenges coupled with GWAS work are further addressed.