ABSTRACT For radiation-related cancer risk evaluation, it is important to assess not only influences of individual risk factors but also their interactive effects (e.g., additive, multiplicative, etc.). Multivariate analysis methods adapted for interactive effects allow such assessments. We have used a multivariate analysis approach to investigate the pair-wise interactions of the previously identified three main etiological factors for lung cancer induction in Russian workers of the Mayak Production Association (PA) nuclear enterprise. These three factors are as follows: (1) body burden of inhaled plutonium-239 (239Pu), an influence on absorbed alpha-radiation dose; (2) cumulative, absorbed external gamma-radiation dose to the lung; and (3) level of cigarette smoking as indicated by a smoking index (SI). The SI represents the cigarettes smoked per day times years smoking. The Mayak PA workers were exposed by inhalation to both soluble and insoluble forms of 239Pu. Based on a cohort of 4,390 persons (77% male), we conducted a nested, case-control study of lung cancer induction using 486 matched cases and controls. Each case was matched to two controls. Matching was based on five factors: sex, year of birth, year work began, profession, and workplace. Three levels of smoking were considered: low (SI = 1 to 499), used as a reference level; middle (SI = 500 to 900); and high (SI = 901 to 2,000). For lung cancer induction, a supra-multiplicative effect was demonstrated for high external gamma-ray doses (> 2.0 Gy) plus high 239Pu intakes (body burden >2.3 kBq). This observation is consistent with the hypothesis of curvilinear dose-response relationships for lung cancer induction by high- and low-LET radiations. The interaction between radiation (external gamma rays or 239Pu body burden) and cigarette smoke was found to depend on the smoking level. For the middle level of smoking in combination with gamma radiation (> 2.0 Gy) or 239Pu body burden (> 2.3 kBq), results were consistent with additive effects. However, for the high level of smoking in combination with gamma radiation (> 2.0 Gy) or 239Pu body burden (> 2.3 kBq), results were consistent with the occurrence of multiplicative effects. These results indicate that low-dose risk estimates for radiation-induced lung cancer derived without adjusting for the influence of cigarette smoking could be greatly overestimated. Further, such systematic error may considerably distort the shape of the risk vs. dose curve and could possibly obscure the presence of a dose threshold for radiation-induced lung cancer.