The objective of the present study was to calculate the gain from selection and compare between the base population (C0) and the best five half – sib families (5% selection intensity). Mass selection cycles C1, and C2 were evaluated for fresh forage yield and yield components.C1 significantly dominated the base population; also C2 significantly dominated C1. The realized gains after the two mass selection cycles were 25.11 , 24.42, 48.41, 72.72, 81.17, 27.46, 21.84 and 73.72% for plant height, leaf /stem ratio, No. of branches/plant, fresh forage yield, dry forage yield, seed number/head, 1000-seed weight and seed yield kg/fad (one Fadden=4200m2), respectively over the base population. The results reveal that the mass selection resulted in high improvement of fresh forage yield, dry forage yield and seed yield, but low gain for plant height, leaf/stem ratio, number of branches /plant, 1000-seed weight and seed No. per head. In general, both genotypic coefficient of variation (GCV) and phenotypic coefficient of variation (PCV) values were low for all studied traits. The values of GCV and PCV were 3.74% and 4.33% for plant height, 6.28 and 9.81% for leaf/stem ratio, 8.08% and 8.38% for No. of branches/plant, 15.39and 16.98% for fresh forage yield, 27.02 and 28.85% for dry forage yield, 11.02 and 11.04% for seed No/head, 8.52and 8.73% for 1000 seed weight and 24.81, 24.83% for seed yield /fad. The heritability estimate increased when the difference between GCV and PCV was low. High estimates of heritability in broad sense were observed for plant height, No of branches/plant, seed No /head, 1000-seed weight and seed yield kg/fad with values of 74.65%, 96.04%, 99.66%, 95.40% and 99.92% respectively. While heritability values for fresh forage yield and dry forage yield were 82.13% and 87.77%, respectively, and was medium for leaf/stem ratio (41.92%).