To investigate the forage yield and yield stability of 11 accessions of Elymus hispidus, a field experiment was laid out in a complete randomized block design with three replications, under two conditions (normal irrigation and dryland farming). The experiment was carried out in two years at the Islam Abad e Gharb research station, Kermanshah, Iran. The collection origin of accessions was Sabalan rangelands in Ardabil Province. Results showed highly significant differences (P<0.01) among the accessions and treatments. In addition, the accession × environment interaction effect was significant (P<0.01) for forage yield. Duncan’s Multiple Range Test (p<0.05) revealed that 890-5, 890-6 and 890-7 were superior accessions and cluster analysis classified them into the same group. AMMI analysis showed that the first significant interaction principal component, explained 97.92% of the total interaction variance. Considering these results, the accessions 890-6, 890-15, 890-5, 890-9, 890-11, and 890-13 showed the highest stability. Stability analysis according to the coefficient of environmental variations (CVi), Wrick’s equivalence (Wi2), stability variance (δi2), coefficients of determination (r2) and regression coefficient (b) revealed that 890-5, 890-6, 890-11 and 890-13 had the highest stability among all accessions while 890-7 had the lowest stability. These results demonstrated that accessions 890-5 and 890-6, which had the best, forage yield and yield stability, were suitable for using in future breeding programs in Kermanshah climate.