The study is based on a case-control study nested within a cohort study (11 000 miners and 1074 lung cancers). The controls were individually matched by year of birth and attained age. Smoking data were collected in person or from relatives of deceased subjects or from medical files. The study resulted in 850 cases of lung cancer with smoking data. The linear dependence of lung cancer relative risk (RR) on radon exposure adjusted for smoking was not substantially different from analyses when smoking was ignored and reflected mainly the risk among smokers. However, the excess RR per unit exposure among never smokers (70 cases) was substantially higher in comparison with that in smokers, reflecting differences in lung morphometry and clearance. The RRs from combined effects are substantially lower than the risk derived from the multiplicative model, but somewhat higher than those from the additive model. The work was supported by the Czech Ministry of Health (IGA NS 10596).