Objectives-Silica has recently been reclassified as carcinogenic to humans based largely on the observed increase in rates of lung cancer in subjects with silicosis. Other recent reviews have arrived at different conclusions as to whether silicosis or silica itself is carcinogenic. This study aims to examine exposure-response relations between exposure to silica and subsequent silicosis and lung cancer in a cohort of goldminers.Methods-2297 goldminers from Kalgoorlie in Western Australia were examined in 1961, 1974, and 1975. Data were collected on respiratory symptoms, smoking habits, and employment history. Subjects were followed up to the end of 1993. Survival analyses for lung cancer mortality and incidence of compensated silicosis were performed with age and year matched conditional logistic regression analyses.Results-89% of the cohort were traced to the end of 1993. 84% of the men had smoked at some time and 66% were current smokers. 1386 deaths occurred during the follow up period, 138 from lung cancer, and 631 subjects were compensated for silicosis. A strong effect of smoking on mortality from lung cancer, and a smaller effect on the incidence of compensated silicosis was found. There was a strong effect of duration and intensity of exposure on the incidence of silicosis. The risk of mortality from lung cancer increased after compensation for silicosis. Of all direct measures of exposure to silica, only log cumulative exposure was significantly related to incidence of lung cancer, but this effect disappeared once the onset of silicosis was taken into account.Conclusions The incidence of silicosis was clearly related to exposure to silica and the onset of silicosis conferred a significant increase in risk for subsequent lung cancer, but there was no evidence that exposure to silica caused lung cancer in the absence of silicosis.