In order to analysis of traits path effective on grain yield of maize hybrid SC.704 under Irrigation-off at growth stages, pattern and plant density in the climate of Ahvaz, an experiment was performed by using split split plots in the form of complete randomized block design with three replications, in the summer between 1386 and 1387 years. Irrigation-off at different growth stages as the main treatment with four levels of irrigation-off in eight leaf stage (S1), irrigation-off in twelve leaf stage (S2), irrigation-off in male florescence appearance stage (S3) and full irrigation (S4) in the main plots, planting pattern with two-level pattern of a row and two rows (zigzag) in the sub plots, and plant density as the sub-sub treatment with three levels of density D1=75000, D2=85000and D3=95000 plants per hectare were placed in sub-sub treatments. Path analysis results showed that the most direct effect on increasing grain yield per unit area, was related to: the number of grain per unit area, with 9.78, the number of grains per corn with 6.03 and thousand grain weight with 3.7, respectively. The indirect effect of high grain number per unit area on grain yield through indirect effects of grain number in a row, the number of grain in corn, biological yield and harvest index to increase grain number per unit area can be justified. The number of grains in corn; also is indirectly effected by grain number in a row, the number of grains per unit area, biological yield and harvest index of the second mentioned indirect effect on grain yield that seems increased biological yield through providing suitable conditions for the entry 0f plant into the reproductive stage affects grain number per unit area and the number of grain per corn, indirectly, has been increased the grain yield per unit area in this experiment. Stepwise regression results and dependent traits showed that 99.86 percent grain yield changes effected by biological yield and harvest index was explained. The role of biological yield with the 97.98 percent ratio on increasing grain yield reflects the effect of this factor on increasing yield grain. Similarity in cluster analysis of grain traits, the number of grains in a row, the number of grains per corn with 99.76 percent and biological yield with grain yield with99.52 percent similarity showed the highest uniformity in changing traits in this experiment.