The authors report the results of the grading of urothelial papillary carcinoma using histomorphometry and pattern recognition analysis. Features concerning nuclear area, the nuclear roundness factor and inclination angle were measured in 38 cases of urothelial papillary carcinomas. The relative frequency distribution showed overlapping among the grades with a reduction in the practical significance of histomorphometry. To overcome this problem, the pattern recognition analysis was applied. It was found that mean nuclear area, mean roundness factor and percentages of round and horizontal nuclei represent the optimal discriminating set of the subsequent steps in the malignant urothelial transformation.