A group of 343 patients with bladder carcinomas was uniformly staged, both clinico-radiologically and pathologically. In accordance with pathological staging, they were treated from 1983 to 1990 and follow-up was closed on January 1992. No systemic chemotherapy regime was used. The present study was designed to assess the value of classical morphological parameters (tumour extension, histological subtype, grade and growth pattern) in the prediction of prognosis, and also to evaluate the adequacy of the current TNM classification (4th edition, 1987) of bladder cancer. The initial tumour stage appears the most useful criterion in the prediction of prognosis. Nevertheless, survival analysis confirms the necessity to modify the present TNM classification for routine clinical practice. In fact, stage III proves to be heterogeneous, and the difference in survival between categories pT3a and pT3b is even more statistically significant (log-rank P < 0.01) than the difference between pT2 and pT3 as a whole (log-rank P < 0.02). Consequently, invasion of the muscular layer should be reclassified into a common stage II, equivalent to the B category in the ABCD system. Moreover, stage IV is also heterogeneous in terms of survival. Despite the overall life-expectancy being rather poor for a patient with bladder carcinoma, three subsets with different prognosis (log-rank P < 0.001) can be identified: pT4N0M0; pTxN1-3M0; pTxNyM1, where x and y represent any number. Therefore, we believe that various subgroups should be distinguished in a future edition of the TNM classification. Current treatment modalities, involving the role of systemic chemotherapy and aimed at bladder preservation, make such innovations even more convenient for a new edition of the TNM classification of bladder cancer. Apart from tumour staging, several microscopic morphological parameters are valuable in distinguishing patients with different prognosis. Pure transitional-cell histology, papillar growth, and low grade, are favourable data. In fact, tumour grade, although somewhat subjective, is a factor of major prognostic importance. Pauwels' distinction of intermedium grade 2 into 2A and 2B is also helpful in the assessment of a population of "intermediate" prognosis. Similarly, with regard to superficial tumours, the division of infiltration levels of subepithelial connective tissue into "superficial" or "deep into the muscularis mucosae", is also relevant, even after stratification by grade.