This investigation was conducted to study drought tolerance in different populations of sainfoin using morphological and physiological traits. Twenty five cultivars were evaluated for 4 harvest according a split plot design with three replication under non-stress and stress moisture environments. Results indicated that water stress decreased genetic variation for most of the traits. In both moisture conditions, the coefficients of genetic variation for forage yield was highest and for the percentage of dry matter, number of days to flowering and chlorophyll content was lowest. Broad-sense heritability was found high for proline content (87%), plant height (87%), leaf relative water percentage (83%) and panicle length (82%). The results showed that drought stress significantly increased the percentage of leaf, leaf to stem ratio and the proline content while forage yield, yield components and other studied attributes considerably decreased. Significant differences were observed between sainfoin cultivars for all traits with the exception of chlorophyll ratio (a/b) and carotenoids content. Among studied accessions, Kabootarabad, Fereidunshahr and Khansar in non-stress condition and Kabootarabad, Fereidunshahr and Janat Abad in moisture stress condition had the highest forage yield. Based on leaf to stem ratio, cultivars Arak and Aqadash (Semirom) in non-stress condition and Arak, Khansar, Aqadash (Semirom) and Juyfort (Fereidunshahr) in stress condition performed the better quality. The highest dry matter yield reduction caused by water stress (33 percent) and the highest increase in leaf to stem ratio (67 percent) occurred in the forth cut. Significant interaction were observed between plant genotype and harvest and also cut and environment. Although proline content was increased more than six times under drought stress condition, it did not have any association with forage yield and drought tolerance indices. Results indicated that the Stress Tolerance Index (STI) and Geometric Mean Productivity (GMP) are effective selection criteria for identification of tolerant genotypes. Based on STI index cultivar Fereidunshahr was the most tolerant cultivar and Arak, Khansar, Borujen, Semirom and Prly, were the most sensitive cultivars. Results of biplot analysis based on the two first principle components showed that cultivar Fereidunshahr, Khorram-Abad, Jannat-Abad, Damaneh and Isfahan had the highest yield potential and the less sensitivity to drought stress. Stepwise regression analysis in non-stress condition showed that the percentage of stem, percentage of dry matter yield and number of stems per meter square justified the majority of dry matter yield variation. Under stress condition the percentage of stem, percentage of dry matter yield and plant height were the most components of forage yield.programs.