Indoor radon measurements were carried out in cave dwellings of the Chinese loess plateau in Gansu province, where previously the Laboratory of Industrial Hygiene (LIH), China, and the U.S. National Cancer Institute (NCI) had conducted an international collaborative epidemiological study. The LIH-NCI study showed an increased lung cancer risk due to high residential radon levels, and estimated the excess odds ratio at 100 Bq/m³ to be 0.19 (Wang et al., 2002). The present study used two types of newly developed passive monitors: One is a discriminative monitor for radon and thoron; the other is a selective monitor for thoron decay products. The arithmetic mean concentrations of indoor radon and thoron were 91 and 351 Bq/m³, respectively. As reported by our previous study in Shanxi and Shaanxi provinces (Tokonami et al., 2004), the presence of high thoron concentration was confirmed and thoron was predominant over radon in the cave dwellings. However, the mean equilibrium equivalent thoron concentration (EETC) was found to be much lower than expected when assuming the equilibrium factor of 0.1 provided by the UNSCEAR (2000) report. The effective dose by radon and thoron decay products was estimated to be 3.08 mSv/yr. It was significantly lower than the dose of 8.22 mSv/yr estimated from the measurements that did not take into consideration any discrimination between radon and thoron. Excess relative risk of lung cancer per sievert may be much higher than the risk estimated from the LIH-NCI study, considering that discriminative measurements were not used in their study.