A study of liver and colon cancers and their association with a previous diagnosis of schistosomiasis was performed in rural Sichuan, China. The data analysed came from 127 liver-cancer and 142 colon-cancer patients, each matched, by age, gender, hospital and township, with one or two controls. The cancer cases were identified from the hospital records of three Sichuan counties (all of which have some level of endemic schistosomiasis japonicum): Meishan (66 liver and 54 colon cases), Pujiang (16 liver and 22 colon cases) and Xichang (45 liver and 66 colon cases). Each control was selected using the hospital records for the same year the matched case of cancer was diagnosed, when the control had been found to have an illness other than cancer. Previous schistosomal infection was determined by examining the medical records at the county hospitals, searching the records at the local schistosomiasis-control stations or health-surveillance units, and, when no written record could be found, by interviews with the subjects or their relatives. Given the extremely strong association between hepatitis and liver cancer in China, only data from hepatitis-negative pairs were used in the analyses. Previous schistosomal infection was found to be significantly associated with both liver cancer (odds ratio = 3.7; 95% confidence interval = 1.0-13) and colon cancer (odds ratio = 3.3; 95% confidence interval = 1.8-6.1). The results indicate a fraction of disease attributable to schistosomiasis of 24% for colon cancer, and (among the hepatitis-negative population) one of 27% for liver cancer.