The manufacturing industry today faces a highly volatile market in which manufacturing systems must be capable of responding rapidly to market changes while fully exploiting existing resources. Reconfigurable manufacturing systems (RMS) are designed for this purpose and are gradually being deployed by many mid-to-large volume manufacturers. The advent of RMS has given rise to a challenging problem, namely, how to economically and efficiently reconfigure a manufacturing system and the reconfigurable hardware within it so that the system can meet new requirements. This paper presents a solution to this problem that models the reconfigurability of a RMS as a network of potential activities and configurations to which a shortest path graph-searching strategy is applied. Two approaches using the algorithm and a genetic algorithm are employed to perform this search for the reconfiguration plan and reconfigured system that best satisfies the new performance goals. This search engine is implemented within an AI-based computer-aided reconfiguration planning (CARP) framework, which is designed to assist manufacturing engineers in making reconfiguration planning decisions. Two planning problems serve as examples to prove the effectiveness of the CARP framework.