The relationships between characteres are established through correlation coeficientes of path analysis, wich provides understanding of cause and effect directs and indirects on the main variable. Itaimed to determine the direct and indirect phenotypic effects of the agronomically important characters related to grain yield of soybean with indeterminate growth habit and genetic dissimilarity between genotypes. The experiment was conducted at the Federal University of Santa Maria, at Frederico Westphalen – RS Campus, in 2013/2014 crop. The experimental design was randomized block with ten genotypes of soybean (FPS Paranapanema RR, BMX Classe RR, FPS Solimões RR, BMX Potência RR, BMX Força RR, BMX Energia RR, BMX Turbo RR, FPS Iguaçu RR, BMX Tornado RR, BMX Alvo RR), organized in three repetitions. It was evaluated yield components. The data were subjected to analysis of phenotypic path analysis, and the significant variables were subject to genetic dissimilarity analysis the method of Mahalanobis, through the UPGMA clustering method, and relative contribution of morphological characters in dissimilarity for method of Singh. The path analysis reveals for the characters thousand grain weight and length of branches direct effects with correlation coefficient of magnitude higher to grain yield, reveals through the length of internode significant indirect effects. The method of Singh reveals that the characters number of branches and nodes in the branches largest contribution to distinguish between genotypes. The Mahalanobis distance reveals formation of three groups.