Dermoscopy is used for the examination of the skin lesions using dermoscopy tool. Dermoscopy is mainly used to evaluate pigmented skin lesions. In this paper Detection and classification system is presented which accepts dermoscopic image containing lesion as input and predicts the class of lesion as benign or malignant. Pre-processing is applied on dermoscopic images and segmentation is performed using semi-automated thresholding, utomated thresholding and k-means clustering based segmentation. Features are extracted from segmented image and given as an input to classifier. Various classifiers used are k-nearest neighbour, support vector machine and feed forward back propagation neural network. This paper also compares performances of classifiers used.