BACKGROUND: The optimal timing of surgery for Wilms' tumor has been debated for many years. OBJECTIVE: To compare the use of immediate nephrectomy and delayed nephrectomy following chemotherapy with vincristine and actinomycin D for treatment of nonmetastatic Wilms' tumor, in terms of postoperative chemotherapy needed and surgical mortality and morbidity. DESIGN AND INTERVENTION: Between October 1991 and March 2001, centers in the UK, Ireland, Norway and Australia participated in the third United Kingdom Children's Cancer Study Group (UKCCSG) Wilms' tumour trial (UKW3). Patients aged between 6 months and 16 years who were newly diagnosed with potentially resectable nonmetastatic renal tumors received immediate nephrectomy, or percutaneous biopsy and delayed nephrectomy following 6 weeks' treatment with 6 doses of vincristine and 2 doses of actinomycin D. Randomization was stratified by center. Both groups of children received postoperative chemotherapy according to tumor stage and histology. Patients with bilateral kidney tumors or whose diagnosis was not certain were not included. OUTCOME MEASURES: The primary end points of the study were event-free survival and overall survival. RESULTS: In the 205 patients randomized (median age 2 years 10 months; 52% male), 42 events occurred during follow-up (range 48 days to 12 years 1 month [median 7 years, 2 months]). There were 19 events (13 deaths) in the immediate nephrectomy group and 23 events (11 deaths) in the delayed surgery group. The results for event-free survival were similar in both treatment arms (79.6% at 5 years) with a hazard ratio of 1.25 (95% CI 0.68-2.3; P = 0.52). For overall survival the 5-year rate was 89% for both groups, and the hazard ratio was 0.84 (CI 0.38-1.88; P = 0.18). Wilms' histology was confirmed in 186 patients. Stage distribution was improved in children with Wilms' histologies receiving delayed surgery compared with immediate surgery (stage I: 65.2% versus 54.3%, stage II: 23.9% versus 14.9%, stage III 9.8% versus 29.8%), and in the delayed surgery group the trend for stage to shift from stage III to stage II was significant (P = 0.008). No tumor ruptures occurred in the 102 patients receiving preoperative chemotherapy, whereas 15 tumor ruptures were recorded in the 103 patients receiving immediate surgery. There were no relapses in the track of the biopsy needle. One patient in each group died of hepatic veno-occlusive disease following treatment with actinomycin D. Approximately 20% of children in the delayed surgery group were spared the use of postoperative doxorubicin or radiotherapy as a result of the improvement in stage. CONCLUSION: Preoperative chemotherapy allowed a shift in stage distribution and hence reduced requirement for postoperative chemotherapy and radiotherapy without affecting event-free and overall survival at 5 years.