In order to study yield stability, replicated yield trials were performed using 19 barley genotypes in randomized complete block design with 4 replications, for 3 years (1994 to 1996) at 8 research stations of temperate regions of Iran. Statistical methods including combined analysis of variance, regression analysis, estimating 14 stability parameters and pattern analysis were applied. Combined analysis of variance showed that the genotypes had specific adaptations with locations. In regression analysis, genotype × environment interaction’s sum of squares included 14.21% of the total sum of squares and the variance of regression residuals was nonhomogenous. Thus the regression model and its parameters were not suitable for stability analysis. The Wrick`s ecovalence, the variance of a genotype’s across environments and the parameter of AMMI1 model were the most suitable parameters and had adequate repeatability. The stability analysis based on these parameters and also pattern analysis showed that the genotype 5 was high yielding, stable and recommendable for all the regions. Genotype 16 was low yielding and adapted to low yielding regions.
)