The relationship between the quantitative features extracted from digitized images of visually normal intermediate cells in cervical intraepithelial neoplasia (CIN) I and II smears and the subsequent follow-up diagnosis was studied. Using the ISPAHAN interactive system of pattern recognition, the average rate of correct classification of progressive or nonprogressive lesions was 80% at the specimen level. It is concluded that additional diagnostic information regarding the probability of progression of CIN I and II lesions may be obtained by studying the microphotometric features of visually normal intermediate cells in the cervical smears. Significant differences were found between intermediate cells close to obviously dysplastic cells and those at a distance. The implication of this finding and the problems involved in the selection of the cells are briefly discussed.