The study was conducted at Fog era, Pawe, Assosa, Gondar and Mai-tsebri in Ethiopia during 2016-2017 with the objective of identifying high yielding, major disease resistance and stable rice varieties for rain fed upland ecosystem. A total of seventeen rain fed upland genotypes including one check were used for the study. The trial was laid out in randomized complete block design with three replication with a plot sized of 7.5 m2 with six rows in each location. Data were analyzed using combined analysis of variance, GGE bi-plot and AMMI. The combined analysis of variance for grain yield, days to maturity, days to heading, panicle length and filled grain per panicle showed significant difference (P=0.01). G7 and G4 showed significant difference than the standard check on grain yield and better resistance to blast disease and gave grain yield advantage of 11 % and 10 %, respectively. The three way interaction of genotypes x environment x years were revealed significant variation (P=0.01) for yield and other agronomic characters. The GGE biplot analysis showed that PCA 1 and PCA 2 described for 51.63 % and 27.31% of GGE sum of squares, respectively for grain yield, explaining a total of 78.95 % variation. In AMMI bi-plot, environments E2, E3 and E4 exerted strong interaction forces while the rest (E1 and E5) did less. Based on GGE bi-plot analysis result, Genotype (G4) and genotype (G7) were recommended for national variety releasing committee to release for future production by considering their high yielding and stability.