The relative contribution of the various characters to the visual selection made by the 5 rice breeders were evaluated by the method of path coefficient analysis (WRIGHT, 1921, 1960). The 50 lines of F6 and F7, one line being chosen at random from each generation of the 25 rice varietal crosses, were planted in a random pattern with 2 replications. Ten characters, measured on a plot basis, were considered ; grain yield, total weight, culm lenght, panicle length, number of panicles, number of grains per panicle, fertility, 500 grains weight, heading date, and lodging resistance. On the other hand, the lines of each plot were scored from 1 to 10 (10 being the best) in the field observations made at maturity by the 5 rice breeders. Then correlation coefflcients between these scores and the various characters were computed. In general, the scores were significantly correlated with fertility, lodging resistance, grain yield, 500 grains weight, and heading date. Other correlation coefiicients were low and inconsistent. Furthermore, path coefficient analysis of the various characters influencing the score were conducted as a means of analyzing the correlation coefficients into direct and indirect components. As the results, more than half of the effects were accounted for in the path coefficient analysis of the 10 characters considered, of which grain yield, lodging resistance, and fertility had strong influences on the score, while 500 grains weight, heading date, and the other characters were relatively unimportant. Therefore, the positive correlations between either 500 grains weight or heading date and the scores seemed to be largely an indirect effect caused by the 'association of the components. Although individual evaluators were affected somewhat differentially, the high correlations were recognized among the 5 breeders for the score.