The purpose of this study was to evaluate the efficacy of low-dose chemotherapy in infants with localised and unresectable neuroblastoma (NB). All consecutive infants with localised NB and no N-myc amplification were eligible in the SFOP-NBL 94 study. Primary tumour was deemed as unresectable according to imaging data showing any risk of immediate resection. Diagnostic procedures and staging were conducted according to INSS recommendations. For children, provided that they had no threatening symptom (i.e. vital risk or dumb-bell NB with neurologic deficit), chemotherapy consisted in low-dose cyclophosphamide (5 mg(-1)kg day(-1) x 5 days) and vincristine (0.05 mg kg(-1) at day 1)-CV and repeated one to three times every 2 weeks until surgical excision can be safely performed. No postoperative treatment was given. Between January 1995 and December 1999, 134 consecutive infants with localised NB were registered in the study, of whom 39 had an unresectable NB without N-myc amplification. Among them 28 had no threatening symptom and received CV according to the protocol. Objective response was observed in 14 (50%) and the other 14 were given second-line chemotherapy because of no response. Surgery was attempted in 38 patients including 14 after CV alone, leading to complete resection in 23. Relapses occurred in four patients all local. Survival and event-free survival were 100 and 90+/-5% with a median follow-up of 55 months (range 33-93). In conclusion primary low-dose chemotherapy without anthracyclines is efficient in about half of the infants presenting with an unresectable NB and no N-myc amplification, allowing excellent survival rates without jeopardising their long-term outcome even for nonresponding patients who received standard regimen.