Tool condition monitoring has found its importance to meet the requirement of quality production in industries. Machined surface is directly affected by the extent of tool wear. Hence, by analyzing the machined surface, the information about the cutting tool condition can be obtained. This paper presents a novel technique for multi-classification of tool wear states using a kernel-based support vector machine (SVM) technique applied on the features extracted from the gray-level co-occurrence matrix (GLCM) of machined surface images. The tool conditions are classified into sharp, semi-dull, and dull tool states by using Gaussian and polynomial kernels. The proposed method is found to be cost-effective and reliable for online tool wear classification. © 2015, Springer-Verlag London.