This study was designed to test the hypothesis that the risk of lung cancer from asbestos exposure is confined to persons with radiographic evidence of pulmonary fibrosis. Occupational and smoking histories were obtained from 271 patients with a confirmed diagnosis of primary lung cancer and 678 referents (279 with other respiratory disease and 399 with cardiac disease). Histories were reviewed blind to assess the timing, duration, and probability of exposure to asbestos. To allow for a lag between asbestos exposure and the development of lung cancer, subjects were classified by the time they had spent in an occupation entailing definite or probable exposure more than 15 years before diagnosis. The presence and extent of fibrosis was assessed blindly from chest radiographs by three readers and scored for small opacities with the ILO 1989 International Classification of Radiographs of the Pneumoconioses. 93 (34.3%) cases had worked in an occupation with definite or probable asbestos exposure compared with 176 (25.8%) referents (crude odds ratio for lung cancer 1.49, 95% CI 1.09-2.04). After adjustment for age, sex, smoking history, and area of referral, the odds ratio (95% CI) was 2.03 (1.00-4.13) in the subgroup of 211 with a median ILO score for small parenchymal opacities of 1/0 or more, and 1.56 (1.02-2.39) in the 738 with a score of 0/1 or less (ie, those without radiological evidence of pulmonary fibrosis). These results suggest that asbestos is associated with lung cancer even in the absence of radiologically apparent pulmonary fibrosis.