Osteosarcoma is the most frequent second primary cancer occurring during the first 20 years following treatment for a solid cancer in childhood. Using a cohort study of children treated for a solid cancer, we investigated the incidence and etiology of osteosarcoma as a second malignant neoplasm after childhood cancer in a cohort and a case-control study. We analysed the relationship between the local dose of radiation and the risk of osteosarcoma, taking into account chemotherapy received. A cohort study of 4,400 3-year survivors of a first solid cancer during childhood diagnosed in France or the United Kingdom, between 1942 and 1986, revealed 32 subsequent osteosarcomas. In a nested case-control study, we matched 32 cases and 160 controls for sex, type of first cancer, age at first cancer and the duration of follow-up. Parameters studied were the incidence of osteosarcoma, the cumulative local dose of irradiation and the cumulative dose of chemotherapy received by cases and controls. The risk of a osteosarcoma was found to be a linear function of the local dose of radiation (excess relative risk per gray=1.8), and was found to increase with the number of moles of electrophilic agents per square meter but not with other drugs. No interaction was noted between radiotherapy and chemotherapy. Bilateral retinoblastoma, Ewing's sarcoma and soft tissue sarcoma were found to render patients susceptible to a higher risk of developing an osteosarcoma as a second malignant neoplasm. We recommend long-term surveillance of patients who were treated during childhood for bilateral retinoblastoma, Ewing's sarcoma, soft tissue sarcoma, as well as other first cancer treated with radiotherapy plus high doses of chemotherapy, without focusing exclusively on the radiation field.