Kahn (2005) found that ethnic heterogeneity reduces the number of deaths caused by natural disasters, a finding that is contrary to theoretical predictions. This paper casts doubt on this finding and uses cross-country data from 1965 to 2008 to conduct a re-estimation. To alleviate omitted variable bias, a legal origin dummy and additional economic variables are incorporated as independent variables. Further, to control for measurement problems, I have included an ethnic fractionalization index and an ethnic polarization index to capture ethnic heterogeneity. The key finding is that ethnic polarization is positively related to number of deaths, while ethnic fractionalization is not. This implies that ethnic polarization increases the level of damage caused by natural disasters, and is a more appropriate measure for ethnic heterogeneity than ethnic fractionalization.