Cumin (Cuminum cyminum L.) is one of the most important cultivable medicinal plants in Iran and today, it is the second most popular spice in the world after black pepper. It is essential to know the relationships between yield and its components in cumin breeding programs. Therefore, different multivariate statistical analysis was performed on plant characters in cumin to determine the relationships between seed yield and seed yield components under field conditions. Positive and significant relationships were found statistically between the umbel number/plant, seed number/plant and biomass in all sowing dates. There were also very high correlation coefficients for plant height and number of branches with seed yield/plant at most of the sowing dates. Stepwise linear multiple regression analysis for each sowing dates indicated that 89% to 97.2% of the total variation in seed yield/plant was explained by the variation of the seed number/plant, umbel number/plant and 1000-seed weight in the different sowing dates. Path analysis revealed that seed number/plant had strong direct effects on seed yield of cumin in all sowing dates and 1000-seed weight in most sowing dates had positive direct effects on seed yield/plant which shows their potentials in breeding programs. There were high and significant correlation coefficients between umbel number/plant and plant height with seed yield/plant that was resulted from greatest indirect effects of mentioned traits through seed number/plant. In accordance with results of this research, it can be concluded that seed number/plant and 1000-seed weight are good selection criteria for improving seed yield in the cumin, meanwhile umbel number/plant and plant height have good potential to improve seed yield through indirect selection in breeding strategies.