Management of debris is a concern after any major disaster. In particular, debris removal after a disaster presents challenges unique to that disaster. Often, the debris removal process takes months or even years to finish. It is likely to be a concern for some time to come since there exists many factors that make it such a costly and complex operation. The cost is mostly arising from the cost of collection and transportation to the disposal sites. The debris collection and transportation routing problem is the subject of this study. The debris collection operation after disasters is a new capacitated arc routing problem (CARP). The uniqueness of this problem is due to the limited access from one section to the other, as a result of the blocked access by debris. Therefore a new constraint, which is developed in this study as access possibility constraint, has been added to the classical CARP. A tabu search meta-heuristics is also proposed to solve the augmented CARP formulation for the debris collection operation problem. Case studies on a test network as well as on realistic instances based on estimates of debris due to likely large-scale natural disaster in the Tokyo Metropolitan Area have also been reported at the end under various scenarios such as with or without intermediate depot as well as single vs. multiple vehicles (groups) operation.