Non-response can bias study results. This study investigated the extent of response bias on selected study variables associated with colorectal cancer (CRC) arising from failure to provide a blood sample. The Ontario Familial Colorectal Cancer Registry collected data with questionnaires on family history, personal history and frequency of food consumption, in addition to obtaining a blood sample. A total of 3447 cases and controls was compared on age, sex, selected socio-economic and dietary factors, and on personal medical history and family history of colorectal cancer, separately for those who provided a blood sample and for those who did not. Findings indicated that case-control comparisons lead to biased odds ratios for associations between CRC and age, sex, fat and fruit consumption, history of elevated triglycerides, and a family history of colorectal cancer. Biased associations have implications for the design and conduct of genetic epidemiological studies.