In order to investigate genetic variation and selection indices for seed yield improvement in sesame, fifteen breeding lines along with five local populations of this crop were evaluated in a randomized complete block design with three replications at the research farm of Isfahan University of Technology in 2009. In this study, the traits of days to emergence, days to flowering and maturity, plant height, branches/plant, capsule length, capsules/plant, seed weight, seeds/capsules, seed yield/plant, seed yield/plot, seed oil content and oil yield/plot were evaluated. The results of analysis of variance showed that there were significant differences among the genotypes for all of the studied traits. Seed yield/plant and seed yield/plot of the genotypes varried from 13.02 to 35 g and 1297 to 3239 kg/ha, respectively. A high broad-sence heritability and low difference between genetic and phenotypic coefficients of variation was found for all of the studied traits, except for days to emergence. The genetic and phenotypic correlation coefficients showed that the traits of capsule length, capsule/plant, seed weight and seeds/capsule highly and positively correlated with seed yield. The result of stepwise regression showed that capsules/plant, seeds/capsule and seed weight were important components of seed yield/plant and explained 99 percent of its variation. Capsules/plant, seed weight and seeds/capsule were also more important component of seed yeild/plot and explained 74 percent of its variation. Based on the results of path analysis, capsules/plant and seed weight had more direct effect on seed yield/plant and seed yield/plot, respectively. Cluster analysis classified the genotypes into two distinct groupes. The highest estimated selection efficiency for genetically improvement of seed yield/plant and seed yield/plot was obtained via selection for seed yield/plant and capsules/plant, respectively. Therefore, capsules/plant can be used as an appropriate selection criterion for improvement of seed yield/plant and seed yield/plot. The Smith-Hazel 1, Smith-Hazel 2 and Pesek-Baker indices were constructed based on seed yield components of capsules/plant, seeds/capsule and seed weight. The results showed that seed yield was highly correlated with each of these indices and the estimated efficiency of indirect selection via these indices was relatively high. Thus, it seems that these selection indices can be effectively used for seed yield improvement. The estimated selection efficiency for seed yield/plot improvement via selection for Pesek-Baker index was more than that of Smith-Hazel ones. However, it seems that efficiency of the Smith-Hazel indices for simultaneous improvement of capsules/plant, seeds/capsule and seed weight is higher than that of Pesek-Baker index.