METHODS: The authors reviewed 131 children enrolled in National Wilms Tumor Study-3 (NWTS-3) who received preoperative treatment for tumors unable to be resected at surgery or judged inoperable by imaging evaluation. Preoperative biopsies were performed on 103 patients. Patients were assigned a pretreatment stage: stage II (11 patients), stage III (39 patients), stage IV (66 patients), and unknown (15 patients). The chemotherapy regimen included dactinomycin and vincristine (81 patients), dactinomycin, vincristine, and doxorubicin (30 patients), dactinomycin, vincristine, doxorubicin, and cyclophosphamide (10 patients), and other (8 patients). Preoperative radiation therapy was started concurrently with chemotherapy (27 patients) or because of lack of response (14 patients). Two patients were given preoperative irradiation without chemotherapy. RESULTS: Response to therapy was assessed after the first trial of chemotherapy. Partial responses were noted in 110 patients (85%), 3 had complete responses, 13 had no response or progression of disease, and 5 patients were not able to be evaluated. There were no significant differences in preoperative response to the different chemotherapy regimens. Median time interval from diagnosis to nephrectomy was 58.5 days. When compared with NWTS-3 patients not receiving preoperative treatment, survival was reduced for patients treated preoperatively (88% vs. 74%, respectively, 4-year survival), which was only partially explained by differences in stage distribution. Median duration of follow-up was 5.9 years. Lack of response to the preoperative treatment was associated with a poor prognosis. Eight children died before removal of the primary tumor. All eight had either progressive disease or no response to the preoperative treatment. CONCLUSIONS: The use of preoperative treatment can facilitate subsequent surgical resection in selected patients with inoperable Wilms tumors. Although these very large tumors--judged unable to be resected--have a somewhat worse prognosis, nephrectomy was completed in 93% of patients after preoperative treatment. However, preoperative treatment will lead to less accurate surgical and pathologic staging, and undertreatment should be avoided in these high-risk patients.