In Madagascar, the promotion of large scale agricultural investments is still high on the political agenda. Hence, at the country level, public debates on business models and inclusiveness remain scarce. Based on qualitative and quantitative assessment, the communication rigorously analyses and quantifies the impacts of large-scale plantation and contract farming schemes. It focuses on one company producing the same crop through large plantation (on a 650 ha domain) and contractual scheme (7 000 farmers). It underlines that large-scale farming model, developed by the company to boost the production of crop A, does not reach its goals. It even proves to be slower to develop and more expensive than the contract farming model. The own estate plantation's low results derive from a land conflict. Despite the company's legal land access that one can have considered as secured, conflicts arise between the company and the farmers but, more seriously, amongst the farmers. On the contrary, contract farming did not awake land tensions. It stimulates market and non market land transactions and, on this basis, seems to contribute to a better efficiency and equity of land distribution. The company generates on its own estate plantation the equivalent of 1 full-time job per hectare, offers income opportunities for 4 times more households and generates about 2,6 times more jobs than in the past situation. But, the companies pays the workers 3 times less than the farmers used to win on their own plots and it does not offer the daily workers to really escape poverty. At last, the land losing households did not endure a forced proletarization but know an exacerbated poverty and the company employment only contribute partly to their reconstruction efforts. One the other hand, the company consolidates (and partly creates) through the contract scheme an equivalent of 1.3 full time jobs per hectare. It does not induce a net increase of agricultural revenue but allows a better remuneration