Among a cohort of 9,279 survivors of childhood neoplasms other than retinoblastoma treated in Britain before 1980, the cumulative risk of a second primary tumor (SPT) by 25 years from 3-year survival was 3.7%. This corresponds to about five times the number expected from rates of cancer occurring in the general population. In the absence of both radiotherapy and chemotherapy, there was four times the expected number of subsequent cancers. The risk of an SPT associated with radiotherapy but not chemotherapy and both radiotherapy and chemotherapy were 6 and 9 times that expected, respectively. There is evidence that radiotherapy was involved in the development of many of the SPT's observed. However, case-control investigations are required to examine the relationship between relative risk of an SPT and therapy in detail. Secondary leukemia appears to occur more frequently among more recently diagnosed children with cancer. It is important to continue to monitor the occurrence of SPT's with a view to identifying the least carcinogenic therapies that are consistent with not compromising survival prospects.