Accurate and efficient identification of failure features of returned used mechanical components/parts is the prerequisite for adaptive remanufacturing. However, due to part-to-part variation, it is error-prone and ad-hoc to manual inspect each part with various defects. This paper proposes a failure feature identification method for adaptive remanufacturing. An innovative identification algorithm is developed to quickly identify the failure features which integrates point-clouds generation, fine-registration and Boolean calculation. For the identified features, hybrid tool path for adaptive remanufacturing can be generated automatically. A turbine blade is taken as an example to demonstrate the efficiency and reliability of the proposed method.