In traditional way, whenever there is any fault in an automotive engine ignition system and change of engine conditions, an automotive mechanic can perform an analysis on the ignition pattern of the engine to check any exceptional symptoms, based on specific domain knowledge (domain features of an ignition pattern). In this paper, support vector machine (SVM) is presented to help solve manual diagnosis problem based on not only the domain features, but also the shape features of patterns captured by using a computer-linked automotive scope meter. Since raw ignition patterns captured by the scope meter are very similar, wavelet packet transform (WPT) is proposed for feature extraction so that the ignition patterns can be comparable under the extracted shape features. However, there exist many redundant points in the extracted features which will degrade the diagnosis performance and speed. Therefore, kernel principal components analysis (KPCA) is employed for dimension reduction. Three different SVMs were constructed for diagnosis and comparison. They were built by (i) using raw captured ignition patterns, (ii) purely extracted pattern features, and (iii) domain features along with extracted pattern features. Experimental results show that linear kernel SVM using domain features and extracted pattern features generates the highest accuracy.