OBJECTIVE--To assess the relative risk of developing a second malignant neoplasm in people with a diagnosis of cancer in childhood and adolescence. DESIGN--Register based follow up study. SETTING--Populations of Nordic countries. SUBJECTS--30,880 people under the age of 20 with a first malignant neoplasm diagnosed during the period 1943-87. MAIN OUTCOME MEASURES--Relative and attributable risks of second malignant neoplasms by type of first cancer, age at first diagnosis, calendar period, sex, and country. Expected figures were based on the appropriate national incidence rates for cancer. RESULTS--247 cases of second malignant neoplasms were observed in 238 patients, yielding a relative risk for cancer of 3.6 (95% confidence interval 3.1 to 4.1). The risk changed significantly from 2.6 in people first diagnosed during the 1940s and 1950s to 6.9 among cohort members included in the late 1970s and 1980s. Increases were observed for most types of cancer. Highest levels of the relative risk were seen during the 10 years immediately after first malignant diagnosis. The incidence of second malignant neoplasms attributable to the first cancer and associated treatments, however, showed a consistent rise throughout the 45 years of follow up. CONCLUSION--The estimated risks for a second malignant neoplasm were significantly lower than those found in most large hospital based studies but compatible with the results from a similar population based study in the United Kingdom. Extent of risk and cancer pattern were similar among the Nordic countries and are believed to be representative for a large part of the European population.