In tissue counter analysis, a novel image analysis tool in the evaluation of dermatoscopic images, a lattice of elementary measuring masks (elements) of equal size is randomly placed over the image and the contents of each element is evaluated by a set of colour and texture features. In this study, we tested the efficiency of tissue counter analysis on the diagnostic discrimination of benign and malignant melanocytic skin lesions. In order to assess the amount of benign and malignant tumour elements in each case, 20 cases each of naevus and malignant melanoma were sampled. Analysis was performed by the Classification and Regression Tree (CART), which provided a subclassification into 32 terminal nodes in the learning sets, 16 of them suggestive for the class 'malignant elements', the remaining 16 for 'benign elements'. For diagnostic assessment, only the percentage of the elements suggestive for malignancy in each lesion was used. ROC analysis showed that a threshold level of 36.45% classified all melanomas and 18 out of 20 naevi correctly (sensitivity 100%, specificity 90%, positive predictive value 90.9%). Prospective studies with larger series of cases will have to be performed for clinical relevance.