Barseem clover is a rapidly growing winter annual that enjoys the largest forage productivity among forages in Egypt. Seed yield is a second product of barseem cultivations. The main objective of the recent study was to evaluate the efficiency of bidirectional selection for seed yield in four population of Meskawi barseem clover along with correlated response in forage yield. As for khadarawi population, the first cycle of selection for high seed yield (C1+) gave a significant increase in seed yield reached 483.2 (g.0.2m-2) over base- population, and 402.5 (g.0.2m-2) over the average of the checks. These values amount to 42.04% over the base population or 32.73% over the average of the check varieties. The first cycle of selection for low seed yield gave insignificant increase in seed yield reached 16.4 (g.0.2m-2) over the base population and 97.7 (g.0.2m-2) over the average of the check varieties. These increments, although, insignificant, amounted to 13.95 and 6.480% relative to the base population and the average of the checks, respectively. It was valuable to notice that, the difference between the first cycle of selection for high seed yield (C1+) and the first cycle of selection for low seed yield (C1-) amounted to 322.8 (g.0.2m-2), which represent superiority of the former by 24.65% over the latter. Regarding Meskawi population, the first cycle of selection for high seed yield significantly surpassed the base population by 559.1(g.0.2m-2), which amount to 57.85% of the obtained seed yield from base population (C0). While, the superiority of the first cycle of selection for high seed yield over the average of the check varieties, although, amounted to 253.3 (g.0.2m-2) which is about 19.91%, was not significant. Also, selection for low seed gave insignificant increase of 202.2 (g.0.2m-2) over the base population, which amount to 20.92%. Meanwhile, the obtained seed yield from the first cycle of selection for low seed yield (C1-), was inferior to the average of the checks by -103.6 (g.0.2m-2) which amount to 8.14%. It was valuable to notice that, the difference between the cycles of bi- directional selection (C1+- C1-) amounted to a significant value of 356.9 (g.0.2m-2) which reached 30.54% over selection for low seed yield. Concerning Helaly population, the first cycle of selection for high seed yield (C1+ ), significantly surpassed the base population by 411.3 (g.0.2m-2) which amount to 34.6% over the base population. Also, that cycle significantly surpassed the average of check varieties by 287.9 (g.0.2m-2) which amount to 21.95% over the average of the checks. He superiority of the first cycle selection for low seed yield (C1-) over the base population or the average of the checks, although, reached 5.89 and - 4.07%, had not reached the level of significance. The difference between bi- directional cycles of selection reached 341.3 (g.0.2m-2), which, represent superiority of (C1+) over (C1-) by 27.13%.Commonly, the correlated response in green forage yield to divergent selection for seed yield was variable among the tested base population. Khadrawi population showed a negative correlated response relative to the base population when selection was practiced for high seed yield. While, in Meskawi population, selection with low seed yield was correlated with an increase in green forage yield. On the other side, in both of Serow and Helally population, a correlated negative response in green forage yield was resulted when selection was practiced for high or low seed yield. The latter responses were not measured as significant.