One of the objectives in rice breeding is to increase grain yield. Grain yield is quantitative characters that are complex and highly influenced by the environment. The selection of superior genotypes should not only base on grain yield, but also need to consider characters that associated with grain yield. This research was to evaluate the relation between grain yield and yield components of experimental hybrid rice. The information would be useful for determining selection criteria in preliminary and advanced yield trial. The experiment was conducted during the rainy season (November-February) of 2013 in Cilacap, Central Java and in Malang, East Java, using Randomized Complete Block Design (RCBD) with four replications. A total of 18 experimental hybrid rice and two check varieties, namely Hipa 8 and Ciherang were used in the study. Results showed that grain yield was affected by locations, genotypes and interaction between locations and genoptypes. The average yield of Hipa 8 was 9 t/ha, while Ciherang was 8.78 t/ha. Panicle length, 1,000 seed weight, percentage of filled grain and number of productive tillers as yield components gave positive effects on the increased of grain yield and could be used as selection criteria for choosing genotype among experimental hybrid rice. There were two experimental hybrids, i.e. A7/BH25B-1B (9.13 t/ha) and A1/CRS516 (9.15 t/ha) that produce the highest grain yield and are potential to be released as new hybrid variety.