Cocksfoot (Dactylis glomerata L.) is a cool-season perennial grass that naturally grows in semi-steppe rangelands in Iran. In recent years, as a result of climate change, coupled with high livestock grazing, rangeland productivity has decreased. This study aimed to analyse the stability of forage dry matter (DM) production of 36 cocksfoot genotypes across four rain-fed environments (Ardebil, Zanjan, Hamadan, and Kermanshah) in Iran. At each location, an experiment was conducted using a randomised complete-block design (RCBD) with three replications over 2 years. The result of combined analysis of variance (ANOVA) across locations showed significant effects of environment (E), genotype (G), and GE interaction (P < 0.01) for DM yield. The E, G and GE interaction effects accounted for 33.19%, 17.98% and 40.52% of the total variance respectively. The GE interaction was subsequently investigated using regression stability, Additive Main effects and Multiplicative Interaction (AMMI), and genotype main effect (G) plus genotype–environment (GE) interaction (GGE) biplot analysis. According to the regression method, genotypes G2, G3, G11 and G12 with a slope close to unity, coupled with high production, had good general stability in all locations. In addition, some genotypes were specifically identified for poor- and high-performing environments. According to the AMMI-2, IPC1 vs IPC2 biplot, genotypes G2, G10, G11, G14 and G15, placed close to the origin of the biplot coupled with higher production, showed general stability in all environments. The best genotypes for the respective environments were also determined. Using the GGE biplot, genotypes G2, G3, G11 and G12 were more stable in all environments. Therefore, on the basis of all analytical methods, three local genotypes, G2 (Karaj), G3 (Marand) and G11 (Qazvin), and a foreign genotype, G14 (from, USA), were identified as most suitable genotypes for breeding improved varieties and cultivation in the study locations and similar areas.