Presentation of clinico-pathological correlation in a series of patients with bladder carcinoma. All of them had a complete pathological and clinical staging following TNM guidelines (UICC 1987). Clinical evaluation consisted of a clinical examination, urography and/or ultrasound, cystoscopy, bimanual palpation under anaesthesia and biopsy. As an option, pelvic CAT, MRI and a bone scan were performed. In all cases a reliable pathological staging was obtained, either from cystectomy or complete TUR. Overall, there is a 66% clinico-pathological correlation (60% for Ta category, 78% for T1, 25% for T2, 57% for T3, and 74% for T4). There is a global error of 34% (40% of cases clinically considered Ta were invasive, 16% T1 were pT2 or more, 42% T2 were pT3 or more, and 10% T3 were pT4; while 6% of those considered T1 were pTa, 33% of T2 were pTa or pT1, 33% of T3 were pT2 or less, and 26% of T4 were pT3 or less). We therefore conclude that when T is lower the risk of being clinically understaged is greater, while higher T values increase the risk of clinical overstaging. From a practical point of view, the most severe errors are in the understaging of T2 and T3 (pT3-pT4) tumours and the overstaging of T2 (pT1) tumours. When cystectomy is performed, the risk of understaging is greater for tumours interpreted as T2-T3 while the risk of overstaging T4 tumours is lower. We conclude that, even when adequate staging of bladder cancer is attempted, pre-treatment tumour classification using the diagnostic methods currently available is far from satisfactory.