A fault diagnostic approach for lithium-ion battery is proposed in this paper, which is suitable for high-power applications such as hybrid electrical vehicle and electromagnetic launch. A serial of abusive experiments including overcharge, over-discharge, and low-temperature operation, which commonly occur during high-power applications, are arranged in order to extract corresponding features. Fault feature vector in the form of electrical parameter variation and incremental capacity analysis curve variation is obtained through reference test. The results from the tests are synthesized to establish the diagnosis system based on fuzzy logic, which provides a detailed description for the fault symptoms and connects them with internal fault mechanisms.