To investigate the possible influence of anticonvulsant treatment on cancer risk, a nested case-control study of 104 lung cancers, 18 bladder cancers, and 322 cancer-free controls was conducted. The background for the study was previous observations among 8004 epileptics in Denmark with a significantly high risk for lung cancer and a significantly low risk for bladder cancer. Cigarette smoking appears to explain the lung cancer excess but not the low risk for bladder cancer, another tobacco-related disease. Information was abstracted on 94 and 95% of the cases and controls, respectively. Lung cancer was not associated with any anticonvulsant drug, but bladder cancer was inversely related to use of phenobarbital (PB). The apparent protective effect of PB was further evaluated in a study of rats given 4-aminobiphenyl (ABP), a bladder carcinogen. The levels of 4-aminobiphenyl adducts in hemoglobin and in bladder and liver DNA were significantly lower in rats given PB prior to 4-aminobiphenyl, compared to controls. These studies suggest that PB may induce drug-metabolizing enzymes of the liver that deactivate bladder carcinogens found in cigarette smoke and provide clues to the role of activation and detoxification of carcinogens in humans.