The physiological quality stability and predictability of seeds from 13 soybean genotypes [Glycine max (L.)Merrill]were analyzed. They were produced at three different locations in the town of Cristalina, Goiás. The field improvement trials were held in the agricultural year of 2004/05. The seed quality evaluation studies took place in the laboratories and in the greenhouse of the Federal University of Viçosa, in a completely randomized design with four repetitions. Germination in paper towels, emergence in sand bed, index of emergence speed and stability analyses were tested through the methods proposed by Lin and Binns (1988)and Annicchiarico (1992). The germination percentage averages, the emergence of plants and the emergence speed index were compared through Tukey
rsquo;s test at 5% probability. In the evaluation of the seeds
rsquo; physiological quality, genotype 7B1454170 was identified as the best, and genotype 9B1459189 as the worst. The genotypes Emgopa 313, 7B1454170, 11B145341 and DM339 were classified as offering high stability in physiological quality, and genotypes 3B1346193 and 9B1459189 offered low predictability. The estimation methods used were efficient, coherent among them and allowed the identification, among the evaluated genotypes, of the ones that offered greater stability and predictability