Long-term results are presented of 28 patients who were diagnosed with neuroblastoma at more than 12 months of age and who received melphalan 180 mg/m2 (n = 6) or 240 mg/m2 (n = 22) to consolidate remissions of Stage IV disease or to control refractory disease. Twenty-four patients also received dianhydrogalactitol 180 to 240 mg/m2, and 11 received total body irradiation 450 to 600 cGy. Autologous bone marrow transplantation (ABMT) was performed with marrow that was unpurged (n = 2) or purged ex vivo (n = 26) with 6-hydroxydopamine (6-OHDA) 20 micrograms/ml plus ascorbate 200 micrograms/ml. The median time to an absolute neutrophil count of 500/microliters was 21 days and to self-sustaining platelet counts more than 20,000/microliters, 28 days. One patient required infusion of unpurged reserve marrow. Two groups of patients underwent ABMT: (1) 17 patients (Group I) who were in first remission a median of 7 months after diagnosis; and (2) 11 patients (Group II) who had refractory disease or were in second remission. For Group I, event-free survival was 29% at 12 months and 6% at 24 months post-ABMT. All Group II patients died of disease or ABMT-related toxicity. Overall, of the 28 patients, one is a long-term relapse-free survivor; five died of ABMT-related toxicity; ten patients with tumors present at ABMT had progressive disease within 6 months of ABMT; and 12 patients with no measurable disease at ABMT relapsed 4 to 32 months (median, 12) post-ABMT. Among the latter, six relapses involved the primary site, and six were restricted to distant sites. These results--in accord with the long-term outcome in other series--suggest that for neuroblastoma high-dose melphalan cannot be relied on to ablate residual disease or to salvage patients with refractory tumors. In addition, the pattern of relapse in several patients could be explained by infusion of incompletely purged autografts; this would support recent laboratory evidence that 6-OHDA/ascorbate is a suboptimal purging method.