Six cropping systems (CS) of three 3-year rotations based on durum wheat and sunflower inserting 0, 1 (pea or fababean) and 2 grain legumes (GL) (pea and soybean), and with or without cover crops (CC) were compared at INRA Toulouse from 2004. This experiment is still on going for a twelfth year. We demonstrated that 6 key points. 1) Pea as a preceding crop increased durum wheat grain production by 8% compared to sunflower as a preceding crop with a mean reduction of N fertilization of 45 kg N ha-1. 2) Inserting GL in the rotations significantly affected the amount of C and N inputs to the soil that were lower than with cereals and consequently led to a decrease in soil organic-C (SOC) and –N contents. 3) N leaching simulated using the STICS model was higher when increasing the number of GL (from 22 to 52 kg N ha-1 after two rotation cycles of 6 years, for 0 to 2 GL respectively). 4) However, CC insertion i) reduced N leaching (from 15 to 18 kg N ha-1), ii) mitigated SOC loss, and iii) did not affect durum wheat grain protein concentration or yield. 5) Daily measured N2O emissions over the whole 3-year rotation were low but significantly higher under the CS including fababean than for the cereal-based CS (1.12 vs. 0.78 kg N2O-N ha-1 year- 1) despite a lower N fertilization. Then, in such conventionally-tilled systems, properly designed cropping systems that simultaneously insert grain legumes and cover crops reduce N requirements of the following durum wheat, stabilize SOC content but do not decrease N2O emissions at the rotation level.