The purpose of this study was to compare self-reports of colorectal cancer (CRC) screening by fecal occult blood test (FOBT), sigmoidoscopy, and colonoscopy with medical records in a multiprovider health care setting. Relatives of CRC patients residing in Ontario, Canada completed a questionnaire indicating whether or not they had ever had any CRC screening tests. Medical records from physician's offices and hospitals were compared with the self reports, and where possible, reasons were obtained for nonmatching reports. Medical records for colonoscopies were readily available from various sources, and self-reports of this procedure were very accurate (kappa statistic for agreement beyond chance = 0.87). For sigmoidoscopy and FOBT, the agreement was poorer (kappa = 0.29 and 0.32, respectively); however, there were difficulties in obtaining records for these two procedures. Sigmoidoscopy procedures that took place many years ago were difficult to document, and physician's offices were unable to provide FOBT reports in many cases. Self-reports of colonoscopy were very accurate in this population, whereas self-reports of sigmoidoscopy and FOBT are somewhat less accurate, although this is likely due to challenges in obtaining a confirmatory record rather than an overreporting of tests. In a multiprovider publicly insured health care setting such as Canada, using self-reported information is likely to provide sufficiently accurate information for colonoscopy, but for other CRC screening tests, there may be difficulty in obtaining true estimates of the frequencies of these procedures.