The Multinomial Logit (MNL) framework has been used in the agricultural production economics literature to model acreage share choices, crop decisions or land use decisions. This article extends the pioneering works of Caswell and Zilberman (1985) and of Wu and Segerson (1995) by developing further the theoretical background of the MNL acreage share models. Two approaches are considered: the “cost function approach” and the “discrete choice approach”. It is then shown that MNL acreage share models can be used to define simple multi-crop econometric models with land as an allocatable fixed input. Finally several generalizations of the standard MNL acreage share model are proposed.