Eighty eight rice genotypes were evaluated during four years aiming to study the relationship among yield components. In each year, the experimental design was a completely randomized block design with four replications and the variables studied were: plant height, seed yield, spikelet sterility, number of seeds per panicle, 1000 seeds weight and head grains. The estimation of correlations and the path coefficient analysis were conducted with the GENES software. The genotypes showed significant variability for all the variables analyzed within each one of years. It was concluded that variable selection must be carried out among the cultivars that are more similar and using a greater number of variables allows better inferences about rice yield components. Grain weight was the yield of component that most affected seed yield rice plants.
path coefficient analysis; variables selection; Oryza sativa L.
