The lung cancer mortality rate has been lower in Japan than in the United States for several decades. We hypothesized that this difference is due to the Japanese preference for cigarettes with charcoal-containing filters, which efficiently absorb selected gas phase components of mainstream smoke including the carcinogen 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone. We analyzed a subset of smokers (396 cases and 545 controls) from a case-control study of lung cancer conducted in Aichi Prefecture, Japan. The risk associated with charcoal filters (73% of all subjects) was evaluated after adjusting for age, sex, education and smoking dose. The odds ratio (OR) associated with charcoal compared with 'plain' cigarette filters was 1.2 (95% confidence intervals [CI] 0.9, 1.6). The histologic-specific risks were similar (e.g. OR = 1.3, 95% CI 0.9, 2.1 for adenocarcinoma). The OR was 1.7 (95% CI 1.1, 2.9) in smokers who switched from 'plain' to charcoal brands. The mean daily number of cigarettes smoked in subjects who switched from 'plain' to charcoal brands was 22.5 and 23.0, respectively. The findings from this study did not indicate that charcoal filters were associated with an attenuated risk of lung cancer. As the detection of a modest benefit or risk (e.g. 10-20%) that can have significant public health impact requires large samples, the findings should be confirmed or refuted in larger studies.