Let us consider a problem to localize a target node located somewhere in a building which is collapsing and burning by a disaster using anchor nodes located outdoors by means of received signal strength (RSS). In this case, the knowledge on the internal architectural structure is required to accurately localize the target node, since the RSSs at the anchor nodes depend on not only the location of the target node but also the layout of walls, doors and windows on floors. However, it is almost impossible to obtain the knowledge in advance. In this paper, we propose a novel indoor localization method. In this method, anchor nodes first estimate the two- or three-dimensional attenuation distribution in terms of wireless signal used for localization, which is called “the attenuation map,” by means of compressed sensing-based wireless tomography and then localize a target node using the estimated attenuation map. We evaluate the performance of the joint attenuation map estimation/indoor localization method by an experiment conducted in an anechoic chamber.