Most of the pattern classification systems employ AI techniques. The most popular one is multi-layer perceptron network (MLP) because of its high computational efficiency. However, there may be some drawbacks: long training time, adjustment of hyperparameters, only a single most probable classification can be returned, etc. In this paper, case-based reasoning (CBR) approach is presented to help solve these drawbacks. One of the advantages of CBR is that multiple possible classifications for a new case can be provided to the user, who can interactively finalize the correct classification. CBR is effective, however inefficient in time because every instance in a case base must be compared during reasoning. To overcome this inefficiency, a clustering technique of kernel K-means (KKM) is employed. To illustrate the effectiveness and efficiency of CBR and clustering framework, an automotive engineering diagnostic problem is shown. Its result is also compared to that of MLP. Experimental results show that CBR even outperforms than MLP. © 2010 IEEE.