Breeding based on selection indices is an effective method for improving complex traits such as yield. To assess the efficiency of different selection indices 28 cultivars of canola (Brassica napus L.) were evaluated under two soil moisture conditions including full irrigation (normal environment) and deficit irrigation (stress environment). The experiment was conducted according to a randomized complete block design with three replications during 2009. Selection indices were computed based on Smith-Hazel and Pesek-Baker methods by seven traits including days to flowering, number of pod per plant, grain number in pod, 1000-grain weight, oil percentage, oil yield and grain yield under both conditions in two forms (with and without yield). Results of genetic correlation coefficients indicated that there was a significant and positive correlation between grain yield and oil yield. Also, the number of pod per plant positively correlated with grain yield and oil yield under both soil moisture conditions indicating that this trait can be useful to indirect improvement of grain and oil yield. In conclusion, results indicated that Smith-Hazel index (with oil yield or grain yield) had the most selection efficiency and can be used to increase oil yield and grain yield in canola breeding programs.