It remains controversial whether silica is a human lung carcinogen. In this study, we estimated the relative risks of lung cancer due to silica and silicosis by meta-analysis. We collected papers published from 1966-2001 which epidemiologically reported on the relationship between silica/silicosis and lung cancer. We removed papers which did not exclude the effects of asbestos and radioactive materials including radon. We selected the most recent one if some papers were based on the same cohort. Based on the selected papers, we summarized the lung cancer risks from silica, silicosis and non-silicosis with exposure to silica, by meta-analysis using a random effects model. The pooled relative risks were 1.32 (95% confidence interval (CI), 1.23-1.41) for silica, 2.37 (95% CI, 1.98-2.84) for silicosis and 0.96 (95% CI, 0.81-1.15) for non-silicosis with exposure to silica. Since some papers on silica did not exclude silicosis, the risk due to silica itself may be smaller than 1.32. It was less possible that silica exposure directly increases lung cancer risk. On the other hand, the relative risk, 2.37 for silicosis suggested that silicosis increases lung cancer risk. Meta-analysis also revealed that cigarette smoking strongly increased the lung cancer risk in silicotic patients (relative risk, 4.47; 95% CI, 3.17-6.30). Thus, the present study suggested the great importance of preventing silicosis and smoking cessation in reducing lung cancer incidence in silica-exposed workers.