PURPOSE: To compare immediate nephroureterectomy with delayed nephroureterectomy after a trial of nephron-sparing endoscopic surgery in patients who were treated initially at our institution from 1996 to 2004 for upper tract urothelial carcinoma. Patients were monitored for upper tract recurrences, metastases, cancer-specific and overall survival. Survival outcomes and perioperative measurements were compared between treatment groups. RESULTS: Of 73 patients, 62 underwent immediate nephroureterectomy and 11 proceeded to nephroureterectomy after failed endoscopic management. Mean follow-up for all patients was 58 months and 75 months for patients who were alive at last follow-up. Patients treated initially with endoscopy averaged a surveillance procedure every 3.7 months and had a median delay to nephroureterectomy of 10 months. Perioperative measurements at time of nephroureterectomy did not differ between groups. Overall survival 5 years from initial resection in the delayed group and from nephroureterectomy in the immediate group was 64% and 59%, respectively; the corresponding 5-year cancer-specific and metastasis-free survival estimates were 91% vs 80% and 77% vs 73%, respectively (P>0.05). Pathologic progression from low to high-grade occurred in three of seven patients from the delayed group. CONCLUSIONS: Failure of endoscopic management necessitating nephroureterectomy does not appear to affect survival outcomes compared with immediate nephroureterectomy in patients with upper tract urothelial carcinoma. A trial of endoscopic management can be considered in patients with low-grade disease and a normal contralateral kidney. Endoscopy is a viable option when there are imperative indications for nephron sparing in the setting of high-grade disease.