We use the 2002 flood disaster in the German state of Saxony as a natural experiment to study whether the population avoided disaster-prone areas after the flood. Such voting-by-feet location choices should enhance the resilience of municipalities in the future. Our difference-in-differences analysis with data from 419 municipalities over more than 10 years, however, shows that the communities affected by the flood had higher migration development than non-affected communities. The differential effect is also economically significant; the net migration rate is higher by approximately 2.5 people per 1000 inhabitants per year in affected municipalities.