In order to study factors which influence the subjective grading of bladder cancers the hypothesis of an ideal examination system was tested. The subjective results were compared with morphometry on the nuclear size of non selected cells. The results show that with precise criteria, with examination of the total section surface field by field and with recording of a grading per field a better interobserver consistency can be obtained. Morphometry demonstrates the heterogeneity and the Gaussian distribution of nuclear sizes in normal and diseased urothelium. It shows that subjective grading is influenced by a small number of larger nuclei and that even in grade 3 tumours 50% of the nuclei have sizes within normal values.