Yield heterosis in rice is reported to interact with environment. Identifying rice hybrids with stable performance is a prerequisite to successfully exploiting this technology over a large area. The popular parameter for selection of stable entries used by many breeders is the varietal mean averaged over locations. Another method is to rank entries for each location and then use mean performance of the rank to assess their potential. It is well known that geno- type environment (G-E) interactions in multilocational yield trials and the comparison of cultivar means are not very useful. In addition, the mean and its ranks do not indicate the location- wise magnitude of differences between the test entry and the standard check against which selection is to be made.