Sectoral policies make explicit and implicit assumptions about the behaviour and capabilities of the agents (such as dynamic responses to market signals, demand-led assistance, collaborative efforts, participation in financing); which we consider to be rather unrealistic. Because of this lack of realism, policies that aim to be neutral often turn out to be highly exclusive. They fail to give sufficient importance to the special features of the sector -with its high climatic, biological and commercial risks and its slow adaptation- or to the fact that those who take decisions in agriculture are now mostly in an inferior position because of their incomes below the poverty line, their inadequate training, their traditions based on centuries of living in precarious conditions, and their geographical location in marginal areas, far from infrastructure and with only a minimum of services and sources of information. These people have only scanty and imperfect access to the markets which, according to the prevailing model, should govern decisions and the (re);distribution of the factors of production. In our opinion, this explains the patchy and lower-than-expected growth registered by the sector after the reforms to promote the liberalization of markets and external openness in the region. In view of the results of the application of the new model, it may be wondered whether Latin America can afford a form of development which excludes over half of its agricultural producers; what the alternatives are; and what costs and benefits each of them offers in terms of production and monetary, social, spatial and other aspects. The article outlines the changes in policies and their results at the aggregate level, summarizes the arguments usually put forward to explain agricultural performance in the region, and proposes a second set of explanations based on a description of the agents and the responses that may be expected from them, contrasting the latter with the supposedly neutral nature of the policies.