From the originally constructed survey data from 2011 Fukushima incident, this paper empirically assessed the sources of failures in disaster risk mitigation in short run. Although residential relocation from the cites at risk is one of the effective risk reduction measures, the relocation incurs mobility costs of developing social capital such as communities or searching public services such as education and medical institutions. The estimation results showed that the residents in the disaster cites of 2011 Fukushima incident can tolerate higher risks of radiation exposure when they have attachment to the original residence or higher demands for the public services, and can stay in the cites at risks consequently. Because the tolerance level can depend on the information associated with the risks, the results imply that the authorities’ providing the correct information is one of the keys for the disaster risk reduction in short run.