Abstract: Abstract：Genotype x environment interaction(GEI)is important in genotype selection and complicates varietal recommendation．Our objective was to compare methods that were selected for lrigh n yield(GY)， n protein yield(GPY)and n protein content(GPC)and stability including gang’s FaBK-suln method(Index 1：equal weights for n yield or quality and shukla’s(1972) stability variance δ2)and five additional rank—sum indices，a stability-variance statistic(s2) and regression coeficient(b)，and to study rank correlations among them and their relationship with GY，GPY and GPC in winter wheat．All statistics were calculated with 27 wheat genotypes grown in 27 environments in France．δ2 and s2 and b statistics would be useful tools for selecting both three grain traits and stability simultaneously．Index1 is slightly more conservative than s2 for GPY．Index 2 selected higher GY and GPC genotypes than did Index 1 and s2．Index 3，Index 4，Index 5 and Index 6 favored selection primarily for GY, GPY and GPC．All the stability statistics were identically well rank correlated with GY，GPY and GPC．The s2 and b statistics were slightly rank-correlated with each other in three grain traits except GY while the high rank-correlation between δ2 and s2 statistics for all three grain traits was detected．The stability statistics between low-yield and high-yield environments were weakly repeatable．Repeatability of the stability statistics was also negligible between statistics estimated in the two years．However．repeatability of all statistics was generally strong between statistics estimated from four randomly chosen subsets of environments．

Key words: genotype x environment interaction, repeatability, stability statistics, winter wheat