A randomized study was designed to determine the effect of intermittent oral cyclophosphamide on the rate of metastases in children with localized and regional neuroblastoma. Secondary objectives of the study were to determine the value of the proposed staging system, to measure the influence of the peripheral lymphocyte count on prognosis, and to study the effect of the primary tumor site and regional node involvement. Since there was no difference in survival in the two groups of patients, whether or not they received chemotherapy, all 113 children were treated as a single group for the analysis of the secondary objectives. The proposed staging served as a good indicator of prognosis with a significant difference in survival seen between each of three stages. It was not possible to detect a significant influence by the three other prognostic variables studied, peripheral lymphocytes, primary site and regional node involvement. Small numbers in the various subgroups studied may account for the failure of differences to achieve statistical significance.