Case-control studies of fatal cancers often rely on proxy respondents. Therefore, it is important to determine the completeness and accuracy of proxy-reported information. We evaluated proxy reports using the Ontario Familial Colon Cancer Registry epidemiology questionnaire. A proxy questionnaire was completed by spouses or relatives identified by a sample of participating cases. Item non-response and percentage agreement (between case and proxy reports) were assessed. More than 30% of proxies were unable to report on physical activity, gynecological surgery, alcohol intake, weight 20 years ago, and oral contraceptive use. Proxy reports of medical history and bowel screening varied, the percentage missing ranging from 5% for diabetes to 44% for familial polyposis in the case of medical history, and from 4% for colonoscopy to 27% for hemoccult tests in the case of screening. Agreement between case and proxy report was good to excellent for colonic screening, most medical history, and for reproductive, medication and vitamin use variables (74% to 100%). It is useful to collect proxy information on such variables as medical history, parity, colonic screening and vitamin use, whereas oral contraceptive use and previous weight are not well reported.