Studies of underground miners have consistently shown an increased risk of lung cancer with cumulative exposure to radon-222 and its decay products. Although the deleterious effects of high radon exposure are clear, questions regarding the shape of the exposure-response relationship, and the effects of time factors such as attained age, time since exposure and early age at first exposure, the effect of exposure rate, and the joint association of radon exposure and tobacco use have not yet been fully clarified. This report considers these questions by fitting various models for the relative odds of disease to 74 male lung cancer cases who were diagnosed between 1981 and 1984 and were alive in 1985 and an equal number of controls. All subjects are current or past employees of the Yunnan Tin Corporation, Gejiu City, China, who reside in the local area. Workers were interviewed to obtain information on work history, from which radon exposure in cumulative working level months and arsenic exposure were estimated, and on tobacco use. Results indicate that excess relative risk increases by 1.7% per cumulative working level month [95% confidence interval (0.5, 5.4)]. The linear exposure response relationship significantly declines with year since last radon exposure (P = 0.02). The risk trend also declines with increasing exposure rate (P = 0.001), indicating that long duration of exposure at a low rate may be more deleterious than short duration of exposure at a high rate. A unique aspect of this study population is the very early ages at first radon exposure for many of the workers, about 37% of the radon-exposed workers were first exposed under the age of 13 years. The analysis shows no modification of the radon lung cancer relationship with age at first exposure. These patterns of risk with radon exposure are generally consistent with those reported in the recent National Academy of Sciences' Biological Effects of Ionizing Radiations IV report. The primary method of tobacco consumption in this area of China is by waterpipe. Lung cancer risk increases with pipe-years of use. The joint analysis of tobacco use and radon exposure supports the Biological Effects of Ionizing Radiations IV conclusion that the most likely model is between additive and multiplicative. The variations of the radon lung cancer relationship by years since last exposure and exposure rate are not affected by adjustment for arsenic exposure.