Seeds of six rice cultivars (Binam, Domsiyah, Shahpasand, Sepidrud, Khazar and Valed 46) along with their diallel crosses progenies were sown in a randomized compelete block design with 3 replications in Rasht in 1990. 10 quantitative traits (grain weight per plant, 1000 seeds weight, no. of filled seeds per panicle, no. of deaf seeds per panicle, panicle length, no. of tillers per plant, plant height, transplantation to first panicle (days), transplantation to full maturity (days) and length to width ratio of rice grain were evaluated and measured on 10 plants. The ANOVA showed significant differences among genotypes for evaluated traits. The genotypic and phenotypic correlation coefficients for each pair of traits were estimated by genotypic variance and covariance analysis. In most cases, the sign of phenotypic and genotypic correlation coefficients were the same. However, in same cases phenotypic and genotypic correlation coefficients were close to each other. This implies that the .environmental variance and covariance were minimal. The genotypic correlation were used for path coefficient analysis. The genotypic correlation coefficient of grain yield per plant with 1000 grain weight (rG=0.31**), no. of filled grains per panicle (rG=0.61**) and no. of tiller per plant (rG=0.37**) were positive and significant. However, genotypic correlation coefficients of grain yield per plant with no. of deaf seeds per panicle (ro=-0.65**), plant height (rG=-0.26**) and days from transplantation to full grain maturity (rG=-0.52**) were significantly negative. The result of path analysis indicated that using traits such as no. of filled grain per panicle and days from transplantation to full grain maturity (earliness of genotypes), can lead the selection for high grain yield per plant. Using traits such as high no. of tillers per plant and short stature of plant, because of their genotypic correlations with grain yield, may be also useful for increasing grain yield per plant in rice.