Despite the five million children who have been screened for neuroblastoma in Japan through detection of catecholamine metabolites, it is still uncertain whether screening for this disease is beneficial. The Japanese study has clearly indicated that screening at 6 months or earlier leads to heavy overdiagnosis. It is shown in this paper that screening at a later age may give the same reduction in mortality with possibly less overdiagnosis. However, it is estimated that, even with two screens at 12 and 18 months, the reduction in mortality would not greatly exceed 25%, under realistic hypotheses on the length of the preclinical phase of the disease. The evaluation of the efficacy of this screening strategy would need the recruitment of half a million children per year over 5-7 years and the follow-up of an equal number of controls. Such a trial would improve our knowledge of the natural history of the disease and might help to answer some questions raised recently regarding its biological heterogeneity.