The utility of automated image analysis in the distinction between poorly differentiated epidermoid carcinoma (eight cases) and small-cell carcinoma (ten cases) was studied. Material obtained using the bronchial brushing technique was prepared by a cytocentrifugation technique. In each case, a total of 100 bronchial cell nuclei were selected using the Leitz TAS, which measured eight parameters per cell in order to ascertain the homogeneity or the heterogeneity of the nuclear populations. Except for one sample exhibiting preparation artifacts, the method proved capable of differentiating between these two types of bronchial carcinoma, with heterogeneity of the malignant nuclei indicating an epidermoid carcinoma and homogeneity indicating a small-cell carcinoma. A correlation was observed to exist between the morphologic and the morphometric criteria.