In this study the linear-bilinear methods were used to understand the complexity of G × E interaction and to evaluate the adaptability and yield stability of some tall fescue genotypes and their selected polycross progenies. In the first step, replicated forage yield data of 72 genotypes (24 parental, 24 early flowering and 24 late flowering progenies) from their evaluation in six main cropping seasons (2008-2014) at two locations and under two levels of irrigation were used for this purpose. The AMMI-1 analysis results accounted for 47.6% of the genotype by environment interaction. Interaction patterns of AMMI-1 biplots indicated that most of the tall fescue genotypes were narrowly adapted, because except for G22, G50, G62 and G65, the rest of the genotypes showed no superior performance at all the tested environments (broad adaptability). Among all the evaluated genotypes, only four (G22, G50, G62 and G65) with yield performance above the average were considered as broadly adapted. In the second step, with focus on 24 parental genotypes through the AMMI model analysis and site regression analysis, the magnitude and significance of the effects of GE interaction and its interaction principal components relative to the main effects of G and E were estimated. The AMMI analysis results accounted for 63.1% of GGE variation, while the SREG GGE biplot analysis results accounted for 73.6%. Although both AMMI and GGE biplot analyses identified G20 as the best genotype, their stability ranking results differed in these analyses. Generally, G20, G14 and G17 were identified as the most stable and high yielding genotypes. In the third step, we used parametric and non-parametric methods. The results indicated that the nine parametric measures of stability identified G12 and G03 as the most stable genotypes. This results for eight non-parametric measures were G15 followed by G11, G03 and G13. Hence these genotypes can be used for improvement of adaptation and stability in tall fescue. Also, principal component analysis based on the rank correlation matrix indicated that most non-parametric stability measures were significantly inter-correlated with parametric measures and therefore, they can be used as alternatives. These results revealed that stability measures can be Keywords: Tall fescue, Genotype × environment interaction, AMMI analysis, GGE biplot, Stability analysis, Parametric and non-parametric methods.