The current work was carried out during 2014/15 and 2015/16 seasons at the Experimental Farm of Agriculture Faculty, South Valley University, Qena Governorate, Egypt to study the direct and indirect effects of yield attributes on forage and seed yields of alfalfa. Six alfalfa genotypes from Egypt (Aswan population, Balady population, El-Dakhla population, Ismalia-1 cultivar, and nitrogen fixing population and Siwa population) beside one variety from USA (Genan) were used in this study. The experiments were laid out in randomized complete block design using split plot arrangement with three replications. Three sowing dates of 20th October (D1), 20th November (D2) and 20th December (D3) were allocated in the main plot while the seven alfalfa genotypes were arranged in the sub plots. Three cuts were taken from each sowing date at 80, 125 and 165 days after sowing at 80, 45- and 40-day intervals, respectively. After taking three cuts, the plants were left out until flowering and seed production which take place in the first week of April, May and June for studied sowing dates, respectively. The obtained results show that, the Ismalia-1 cultivar exceeded the other tested genotypes for seasonal fresh forage yield trait (6.16 kg m-2) under third planting date (20th December) while, El-Dakhla genotype superior with regard to seasonal dry forage yield (2.00 kg m-2) under the same planting date. Otherwise, Aswan population produced the maximum mean values of Seasonal protein forage yield (0.60 kg m-2) under second planting date (20th November). In addition, Genan cultivar which was introduced from USA gave the maximum seed yield plant (1.20 g) under the first planting date (20th October). Furthermore, the obtained results show that the fresh forage yield had the greatest influence on protein forage yield in each sowing date. Meantime, the results of path analysis show that, number of seeds/pod and number of pods/plant considered the most effective traits in seed yield/plant of alfalfa. Moreover, negative correlation between seed yield/plant and 1000-seed weight was observed. Therefore, selection for improving seed yield/plant may be carried directly through selection for number of seeds/pod and number of pods/plant.