To determine whether extracapsular extension of pelvic lymph node metastases from urothelial carcinoma of the bladder is of prognostic significance. From a consecutive series of 507 patients with urothelial carcinoma of the bladder preoperatively staged N0M0, 101 of 124 patients with lymph node metastases detected on histologic examination fulfilled the inclusion criteria for this study and were evaluated. All underwent radical cystectomy between 1985 and 2000 with standardized extended bilateral pelvic lymphadenectomy in curative intent and were prospectively followed for recurrence-free (RFS) and overall (OS) survival. Staging was done according to UICC 2002. A total of 2375 lymph nodes were examined. The median number of nodes examined per patient was 22 (range, 10-43). The median number of positive nodes was 2 (range, 1-24). Median RFS and OS were 17 and 21 months (range for both, 1-191), respectively. The 5-year RFS and OS rates were 32% and 30%, respectively. There were 59 patients (58%) with extracapsular extension of lymph node metastases. They had a significantly decreased RFS (median, 12 vs. 60 months, P=0.0003) and OS (median, 16 vs. 60 months, P <0.0001) compared with those with intranodal metastases. There were no significant differences in survival between pN1 and pN2 categories with extracapsular extension of the lymph node metastases (RFS, P=0.70; OS, P=0.65) or those without extension (RFS, P=0.47; OS, P=0.34). On a multivariate analysis, extracapsular extension of lymph node metastases was the strongest negative predictor for RFS. Meticulous lymph node resection and subsequent thorough histologic examination in patients undergoing radical cystectomy for bladder cancer reveals a high incidence of lymph node-positive disease (24%) despite negative preoperative staging. Lymph node metastases with extracapsular extension in pN1 and pN2 stages carry a very poor prognosis. Therefore, this feature should be used to designate a separate pN category in the staging system. The discrimination of pN1/pN2 in the UICC 2002 classification seems to be arbitrary and of no significant prognostic relevance.