The aim of this study was to classify prospectively a series of neuroblastoma tumours according to the International Neuroblastoma Staging System (INSS) and the International Neuroblastoma Response Criteria (INRC) and to evaluate the difficulties and pitfalls involved in a multicentre setting. Each hospital provided their data for central review. The surgical procedures and their complications were reported. Kaplan-Meier estimates of survival and event-free survival were calculated according to stage and response to therapy. From June 1992 to December 1996, 194 patients were included in the study, with a mean age of 2 years. Initial studies were performed according to INSS recommendations without major problems. INSS stage was correctly applied to all patients except for 9 (95%). Post-operative complications were observed in 15 patients (8.3%). Response to therapy (INRC) was studied in 63 stage 4 patients, 11 of whom were not classified correctly (17%). Differences in survival according to stage (INSS) and group of response to therapy (INRC) were statistically significant (P < 0.001). In conclusion the INSS was easy to use and separated different prognostic groups. Surgical complications and mortality did not increase in this series because of using the INSS. The feasibility of INRC was evaluated in a small series of stage 4 patients and the designation of response was problematic in a relatively high proportion of cases. The prognostic value of the different responses was highly significant, but less informative than had been hoped for.