In this paper, a novel approach is proposed for recognizing pain expression. First of all, supervised locality preserving projections (SLPP) is adopted for extracting feature of pain expression, which can solve the problem that LPP ignores the within-class local structure using adopting prior class label information, and then multiple kernels support vector machines (MKSVM) is employed for recognizing pain expression, Compared to SVM, which can improve the interpretability of decision function and classifier performance. Experimental results are shown to demonstrate the effectiveness of the proposed method.