This experiment was conducted in experimental field at Arsanjan Islamic Azad University during 2007-2008 growing season. In this study a randomized complete block design (RCBD) with four replications was used. Four levels of density (200, 300, 400 and 500 plants per square meter) and five varieties (Darab, Niknezhad, Marvdasht, Chamran and Shiraz) were used. Densities and five varieties showed significant differences for grain yield but their interaction was not significant. 500 plant per m2 had the highest grain yield among different densities (5678 kgh-1) and also Marvdasht had the highest grain yield among different cultivars (4976 kgh-1).Simple correlations between grain yield and its components (number of grains per spike, number of spikes per square meter and TKW) were calculated at 500 plant per m2 density. Number of spikes per square meter and number of grains per spike had significant positive correlation with grain yield, but TKW had negative correlation with grain yield. Simple correlations were divided to direct and indirect effects using Dewey and Lu method. path analysis showed that TKW had a positive direct effect on grain yield, but it had negative indirect effects via number of grains per spike and number of spikes per square meter, which caused a negative simple correlation between TKW and grain yield.According to the results of this study, in breeding bread wheat programs, in order to improve grain yield it is better to pay more attention to the number of grains per spike, and number of spikes per m2 than TKW.