The experiments were conducted using 13 rice genotypes from International Rice Research Institute (IRRI) and commonly grown variety (SARO 5) as a check during April-June 2012 growing season at three rice irrigation schemes of Zanzibar. The aim was to estimate genetic variability, heritability and interrelationships among yield components of genotypes so as to identify potential lines for improvement and production. RCBD with three replications at each site was used. ANOVA showed significant differences for all studied characters indicating that genotypes constitute a lot of genetic variability. GCV were lower than PCV in all studied characters indicating considerable influence of environment on expression of the traits. Broad sense heritability was higher for panicle length, number of tillers per plant, plant height, 1000 grains weight, panicle weight and yield per hectare indicating heritable portion of variation; this suggesting that selection can be carried out on the basis of their phenotypic expression. The correlations that were consistently significant at all locations and in addition to the combined analysis were days to 50%flowering with days to maturity, number of primary branches per panicle with days to 50% flowering, and grain yield per plant with yield/ha suggesting that grain yield per plant should be considered during selection for yield. Path analysis revealed that number of tillers per plant showed positive direct effect on grain yield also had indirect effect via 1000 grains weight and panicle length. Number of grains per panicle had positive direct effect on grain yield also had indirect effect on yield via number of primary branches per panicle. Panicle length showed positive direct effect on grain yield also had indirect effect through number of primary branches. With respect to yield genotype IR07M101 was stable and could be selected for production and further improvement.Commission of Science and Technology (COSTECH)