In this paper, a unique post-earthquake survey designed to provide a rapid assessment of food insecurity in Haiti is used in order to see how adequately emergency assistance programs have been allocated. When modelling the impact of various covariates upon assistance allocation, the location of households emerges as the main criterion. This helps to explain why, five months after the quake, government and agencies still seemed unable to provide an efficient allocation of emergency assistance. What is more, those who benefited less from assistance appeared to be on the one hand families headed by women and on the other hand households with disabled members: this obviously runs counter to an "optimal" targeting that would make the most vulnerable ones eligible for assistance in priority. Furthermore, the fact that associations may favour assistance allocation is an interesting result that should be considered further. It is also found that asset losses had no significant impact on the food consumption score, whereas household pre-earthquake wealth did. This result demonstrates that the impact of the shock has been buffered when households had previously enforced coping strategies, regardless of the effects of emergency assistance programs.