Although radon is a well-established lung carcinogen, there are uncertainties concerning the exposure-response relationship, whether exposures in early life are particularly hazardous, and how smoking affects the risks associated with radon exposure. A cohort study of the mortality experience of 1,743 underground fluorspar miners and 321 surface workers from 1950 to 1984 reported previously has been extended to include 6 additional years of follow-up (1985-1990). A statistically significant relationship was noted between radon-progeny exposure and risk of lung cancer mortality. Our analysis found no effect for age at first exposure. Attained age was strongly predictive of excess relative risk (ERR) per working level month (WLM) of radon exposure, falling from 0.025 for those aged less than 50 years to 0.002 for those 70 years or older. An inverse exposure-rate effect was also observed, wherein for equal total exposure, a high exposure rate (and short duration) is less harmful than a low exposure rate (and long duration). The ERR/WLM increased from 0.0019 for exposures of less than 10 years to 0.0076 for exposures of 20 or more years. The analysis of time-since-exposure windows revealed a greater ERR/WLM for exposures received in more recent periods, similar to the result for time since last exposure. Excess relative risks per WLM were higher for current smokers than for nonsmokers (never and former smokers). Analyses were consistent with a multiplicative relationship between radon-progeny exposure and current smoking and the risk of lung cancer. The assessment of radon exposure and lung cancer risk should incorporate the effects of exposure rate, time since exposure, smoking status and attained age.