A case-control study on interaction among the risk factors of primary lung cancer with 390 matched pairs was carried out in Guangdong from 1990 to 1993. The subjects of study were the in-patients with primary lung cancer (age 32-78), which consisted of 171 cases of primary lung squamous cell cancer (L. S), 138 cases of primary lung adenocarcinoma (L. A), and 81 cases of other pathological type (L. S:L. A = 1:0.8). Primary lung squamous cell cancer accounted for the majority (51.55%) of 291 male cases (L. S:L. A = 1:0.5). Primary lung adenocarcinoma made up the majority (57.57%) of 99 female cases (L. S:L. A = 1:2.7). Single factor conditional logistic regression analysis showed that the occurrence of lung cancer was closely associated with history of chronic bronchitis, emphysema, pulmonary tuberculosis and other pulmonary diseases (OR = 2.9-3.6, chi 2 = 13.52-37.55, P < 0.01) other risk factors were smoking, passive smoking from spouse or in the working place and family history of tumour (OR = 2.7-3.6, chi 2 = 8.53-33.15, P < 0.01). Driver, taking oral contraceptive, liking pickles or salted fish and bad ventilation in kitchen could be the risk factors of lung cancer (OR = 1.3-3.0, chi 2 = 4.78-5.0, P < 0.05). Further multiple conditional logistic regression analysis turned out that history of chronic bronchitis, emphysema, pulmonary tuberculosis, smoking, family history of tumour, passive smoking from spouse or in the working place, liking pickles were the independent risk factors of lung cancer (OR = 1.7-3.5). In search of the interaction between lung cancer and the risk factors which included history of chronic bronchitis, emphysema, pulmonary tuberculosis, smoking and passive smoking, a loglinear model analysis was performed. The results revealed that there was significant interaction between two factors in one-stage analysis. Although the interaction existed between chronic bronchitis and tuberculosis, chronic bronchitis and smoking or passive smoking, tuberculosis and passive smoking in the meantime, it had not attained the level of significance among three or more factors (including the principal effect of lung cancer) in two-stage analysis (chi 2 = 1.31-2.27, P > 0.13), and therefore the interaction of the former was not affected by the effects of the latter as yet.