Ten soybean genotypes representing improved varieties and promising lines were grown on wetland after rice at Muara Experimental Farm, Bogor, West Java, during the 2010 dry season. The experiment was arranged in a randomized block design with three replications. Each of the genotypes was grown in a 2 m x 4.5 m plot at a 40 cm x 15 cm plant spacing, two plants per hill. Results of the data analyses showed that among the yield components, plant height, number of pods per plant, and harvest index were positively correlated with grain yield. The direct effects of plant height, number of pods per plant, and harvest index on grain yield as indicated by the path coefficient were the highest, while the effect of other yield components were either small or negative. The yield variation (IR2) not attributable to the ten yield component variables was high (53.66%). Based on the analyses, soybean genotypes with high grain yield should have sufficient plant height, high number of pods per plant, and high harvest index. Therefore, plant height, number of pods per plant, and harvest index could be used as criteria for selection of high yielding genotypes in the soybean breeding program.