This paper analyses empirically the impact of food aid on wheat production, sales and purchases in rural Ethiopia between 1994 and 2009. We distinguish between the impact at the intensive margin (on quantities) and at the extensive margin (on the very decision to produce or go to the market to buy or sell). The panel dimension allows us to deal with food aid selection. We find that the impact of food aid goes through the extensive margin while pure quantity effects, once controlled for market participation and production choice, are not significant. Food aid reduces the probability of being a producer albeit the size of the effect is small. It also increases the probability of being a seller and decreases the probability of being a buyer after 2004, the year when the rules of food aid allocation changed in Ethiopia. Other factors such as storage capacity, distance to the nearest market, and the frequency the market is held also matter in the decision to sell or buy.