The purpose of our study was to investigate the interrelationship of known and possible new risk factors in patients with metastatic neuroblastoma and to define groups at risk. The possible influence of 37 variables on event-free survival (EFS) was analyzed univariately in 308 consecutive patients using the Kaplan-Meier estimate. Fifteen factors were identified (p less than 0.05, logrank greater than 3.84) of whom eight showed a nonrandom correlation to several others (chi 2-test, p less than 0.05). Seven noncorrelated factors [lactate dehydrogenase (LDH) level, resectability of the primary tumor, histologic grade, leukopenia, presence of symptoms, general condition, and age at diagnosis] were included in the multivariate analysis of 182 patients according to the Cox model. The variables LDH (p = 0.0007), resectability (p = 0.0063), histologic grade (p = 0.0055), and leukopenia (p = 0.0470) were identified multivariately as prognostic factors for EFS. These results permitted the classification of patients into three prognostic groups. The 6 year event-free survival for group IV-A (LDH normal) was 0.37 +/- 0.12, for group IV-B (LDH abnormal, additional risk factors favorable) 0.18 +/- 0.10, and for group IV-C (LDH abnormal, 1-3 additional risk factors unfavorable) was 0.08 +/- 0.03. We conclude that the proposed clinicopathological classification may prove to be a reliable and easily applicable tool for estimating the outcome of metastatic neuroblastoma in children.