Bone sarcoma as a second malignancy is rare but highly fatal. The present knowledge about radiation-absorbed organ dose-response is insufficient to predict the risks induced by radiation therapy techniques. The objective of the present study was to assess the treatment-induced risk for bone sarcoma following a childhood cancer and particularly the related risk of radiotherapy. Therefore, a retrospective cohort of 4,171 survivors of a solid childhood cancer treated between 1942 and 1986 in France and Britain has been followed prospectively. We collected detailed information on treatments received during childhood cancer. Additionally, an innovative methodology has been developed to evaluate the dose-response relationship between bone sarcoma and radiation dose throughout this cohort. The median follow-up was 26 years, and 39 patients had developed bone sarcoma. It was found that the overall incidence was 45-fold higher [standardized incidence ratio 44.8, 95 % confidence interval (CI) 31.0-59.8] than expected from the general population, and the absolute excess risk was 35.1 per 100,000 person-years (95 % CI 24.0-47.1). The risk of bone sarcoma increased slowly up to a cumulative radiation organ absorbed dose of 15 Gy [hazard ratio (HR) = 8.2, 95 % CI 1.6-42.9] and then strongly increased for higher radiation doses (HR for 30 Gy or more 117.9, 95 % CI 36.5-380.6), compared with patients not treated with radiotherapy. A linear model with an excess relative risk per Gy of 1.77 (95 % CI 0.6213-5.935) provided a close fit to the data. These findings have important therapeutic implications: Lowering the radiation dose to the bones should reduce the incidence of secondary bone sarcomas. Other therapeutic solutions should be preferred to radiotherapy in bone sarcoma-sensitive areas.