Superoxide dismutases (SODs) are antioxidant enzymes that play a role in the defense system of the body. They may be involved in protection against carcinogenesis processes. In the present study, we investigate the association between serum SOD activity and the risk of deaths due to all cancers combined, based on a nested case-control study within the Japan Collaborative Cohort Study of 914 cancer deaths and 2,739 matched controls. Blood samples were obtained at the baseline and stored at -80 degrees C until analysis for SOD levels. Serum levels of SODs were divided into quartiles, with the first quartile used as the reference. A conditional logistic model was used to estimate odds ratios (ORs) for total cancer mortality associated with serum SOD quartile levels. The adjusted ORs and 95% confidence intervals (CIs) for the second, third and fourth SOD quartiles were 0.96 (95%CI: 0.77-1.19), 1.18 (0.92-1.51), and 1.32 (1.04-1.69), respectively. In analyses stratified by observation period, the adjusted ORs of the respective quartiles were 0.81 (95%CI: 0.60-1.08), 0.98 (0.70-1.37), and 1.28 (0.92-1.79) for the period from the baseline to 1994; and the adjusted ORs were 1.18 (95%CI: 0.85-1.63), 1.47 (1.04-2.10), and 1.41 (1.00-2.04) for the period after 1994. To conclude, we found a slightly positive association between serum SOD level and the risk of all cancer mortality in the present study. Elevated serum SOD levels might reflect a response to oxidative stress, and then may predict a state of excess reactive oxygen species in the carcinogenesis process. Detailed studies of associations between serum SOD levels and cancers in specific sites should now be performed, with attention to particular tumour types.