OBJECTIVE Exposure to the radioactive daughters of radon is associated with increased risk of lung cancer in mining populations. An investigation of incidence of lung cancer following a clinical survey of Ontario uranium miners was undertaken to explore whether risk associated with radon is modified by factors including smoking, radiographic silicosis, clinical symptoms, the results of lung function testing, and the temporal pattern of radon exposure. METHODS Miners were examined in 1974 by a respiratory questionnaire, tests of lung function, and chest radiography. A random selection of 733 (75%) of the original 973 participants was followed up by linkage to the Ontario Mortality and Cancer Registries. RESULTS Incidence of lung cancer was increased threefold. Risk of lung cancer among miners who had stopped smoking was half that of men who continued to smoke. There was no interaction between smoking and radon exposure. Men with lung function test results consistent with airways obstruction had an increased risk of lung cancer, even after adjustment for cigarette smoking. There was no association between radiographic silicosis and risk of lung cancer. Lung cancer was associated with exposures to radon daughters accumulated in a time window four to 14 years before diagnosis, but there was little association with exposures incurred earlier than 14 years before diagnosis. Among the men diagnosed with lung cancer, the mean and median dose rates were 2.6 working level months (WLM) a year and 1.8 WLM/year in the four to 14 year exposure window. CONCLUSIONS Risk of lung cancer associated with radon is modified by dose and time from exposure. Risk can be substantially decreased by stopping smoking.