Four major staging systems have been used to estimate the prognosis for children with local and regional neuroblastoma (NBL). Data obtained at diagnosis for 251 neuroblastoma patients from two Childrens Cancer Study Group (CCSG) studies were analyzed according to staging systems of the CCSG, St Jude Children's Research Hospital, the Pediatric Oncology Group (POG), and the Union Internationale Contre le Cancer (UICC) tumor-nodes-metastasis (TNM) system. The most significant variables were found to be age, tumor stage, extent of tumor removal, transgression of the midline by tumor infiltration, and site of primary tumor. Involvement of lymph nodes per se was not a bad prognostic sign unless associated with extension beyond the midline, the latter being the single most important prognostic variable. All four staging systems had value for prognostication and all identified with accuracy the low stage patient (stage I, stage A) who fares well (greater than or equal to 87% survival). The CCSG definition of stages II and III disease discriminated prognostic groups best among the remaining patients, and was able to identify the child with local-regional NBL with poor survival. The estimated 5-year survival rates for children with regional tumor (stage III, IIIA[N]), according to the four systems were 44%, 74%, 74%, and 74% for the CCSG, St Jude, POG, and UICC methods, respectively. We conclude that all four staging systems effectively define good-prognosis patients with localized disease but that the CCSG staging system most accurately identifies patients with regional tumor who have a poor outcome.