The classification of Tumors is crucial to the proper treatment of cancers. In recent years, sparse representation-based classifier (SRC) has been shown its good classification performance. It has been successfully used for classification of tumors using gene expression data. However SCR could not well classify the data with the same direction distribution which is also existed in the gene expression data. This paper presents a new method, metagenes-based kernel sparse representation for tumors classification (MKSRC). To make the data in an input space separable, we implicitly map these data into a kernel feature space by using some kernel functions. A set of metagenes which can capture the structures inherent to the data are extracted from the training samples and are more effective than the original gene expression data for classification. Extensive experiments on publicly available gene expression data sets show that the performance of MKSRC is comparable with or better than many existing representative methods.