There is little micro-evidence on the persistence of natural disasters' welfare impacts. This paper assesses the effect of Hurricane Mitch on consumption of Nicaraguan agricultura) households. Mitch occurred in October 1998. Pre-post data is obtained from a nationally representative panel collected in 1998 and 2001. An additional survey was fielded in 1999 for households from the panel affected by the hurricane. The 1999 data contains self-reported measures of hurricane-induced losses. Satellite rainfall observations are used as a complementary measure of the shock. Using the structure of the data, the paper disentangles the idiosyncratic and common dimensions of the shock, together with its short and mediumterm impacts. Within the sample of households affected by the hurricane, micro-growth model estimates point to short-term negative effects at most limited to idiosyncratic events such as floods and displacement. Mitch 's medium-term common impact is then analyzed as a quasiexperiment. Difference-in-differences estimates do not provide evidence that households affected by Mitch suffered from lower growth between 1998 and 2001. Overall, hurricane Mitch's direct consumption impact thus exhibits little persistence.