Cancer mortality was studied in 10,552 Swedish hyperthyroid patients treated with '"1 between 1950 and 1975. The patients were matched with the Swedish Cause-of-Death Register and the cases of 977 patients who died from cancer or leukemia were studied. The patients had been followed up for an average of I5 years (range 0 to 35 years), and the overall standardized mortality ratio (SMR) was 1.09 [95% confidence interval (CI) = I .03 to I. 161, with a higher risk for women. The highest mortality was seen during the first year after exposure (SMR = 1.15) and decreased for the following 9 years (SMR = 1-04). The risk of dying from a cancer in the digestive tract and respiratory organs was significantly elevated more than I0 years after exposure, as was the overall cancer mortality (SMR = 1.14). No increased risk was seen for leukemia, bladder cancer or breast cancer. Younger patients and those receiving "'I at higher activity had higher SMRs than older patients and those receiving lower activity. Patients with toxic nodular goiter had higher risk than those with Graves' disease. The lack of increasing mortality over time and with increasing activity of '"I administered argues against a carcinogenic effect of I3'l. However, in the case of cancers of the stomach, the "'I exposure could have contributed to the excess mortality from these cancers.