This paper is the first to uncover in details the impact of different families of disasters on exports from 1979 to 2000 (storms, floods, earthquakes and changes in temperatures). Besides, our paper is the first to compare in a quasi-systematic way the results across the two datasets at hand, the standard \emph{EM-DAT} data and \emph{GeoMet} data, a newly available dataset based on geophysical and meteorological data (\citet{Felbermayr2013} and \citet{Felbermayr2014}). We run series of regressions while accounting progressively for the characteristics of products (all traded goods v/s agriculture ones), the characteristics of the country (size, level of development) and the intensity of the catastrophes. When pooling all countries, and all types of disasters, we do not find any statistical impact on exports. But when focusing on each of them separately and on agricultural goods, the occurrence of an earthquake appears to reduce exports of about 3\%, regardless of its location. A windstorm shock, even when it happens to be very severe, has hardly any impact. A flood, on its side, is estimated to reduce export flows of a small country by nearly 3\%. The effect of changes in temperatures is ambiguous. All in all, except for temperatures related disasters, the results are consistent across both datasets, EM-DAT and GeoMet, although they appear to be slightly more in line with our expectations in the case of GeoMet.