Purpose: The aim of this study was to evaluate the accuracy of frozen section analysis (FSA) for detecting and eliminating malignant ureteral margins at radical cystectomy (RC) and to determine the impact of final margin status obtained by a sequential resectioning strategy on the risk of upper urinary tract recurrence (UUT-R). Methods: A total of 425 ureteral margins obtained from 218 patients undergoing RC for invasive bladder cancer between 1999 and 2009 were initially examined by FSA. When positive, additional resections were performed. Subsequently, all specimens were reexamined on formalin-fixed, paraffin-embedded (FFPE) sections. Fisher-Exact test/logistic regression were used for uni-/multivariate analysis and log-rank test for the impact of margin status on UUT-R. Results: Of 425 specimens, malignant ureteral margins were found on initial FSA in 17, on FFPE in 23 (sensitivity: 73.9%). FSA results were false-positive in 1/402 margins (specificity: 99.8%) resulting in an overall accuracy of 98.3%. On multivariate analysis (95%-CI), correlations were found between distal ureteral malignancy and FSA (p < 0.0001) and tumor multifocality (p = 0.04). In 10/17 positive initial margins it was not possible to obtain a negative final margin despite multiple reresections, of which one resulted in a UUT recurrence, which was more frequent than in patients with a negative margin (4/208, p = 0.03). Moreover, three of these five recurrences were proximally to the anastomosis. All patients had multifocal disease at RC. Conclusions: FSA has a high accuracy for detecting malignant ureteral margins. Patients with positive final margins are at increased risk of UUT-R. With sequential resection, however, positive margins cannot reliably be converted to negative ones.