This article characterises vulnerability to poverty in Haiti using a unique survey conducted in 2007 in rural areas. In a first step, using two-level linear random coefficient models of both per capita consumption and per capita income, the article assesses the impact of self-reported shocks on households' economic well-being. In a second step, the prediction model is used to calculate various measures of vulnerability to poverty, considering various types of shocks. Empirical findings show that self-reported (or observable) idiosyncratic shocks, in particular health-related shocks, have larger impact on vulnerability to poverty than observable covariate shocks. These results are in line with the fact that many households reported idiosyncratic health shocks as being the worst shocks they experienced. On the other hand, unobservable idiosyncratic shocks appear to have generally more influence on households' vulnerability to poverty than unobservable covariate ones. We also show that omitting self-reported shocks in the analysis leads to an underestimate of households' vulnerability to poverty.