Background: Recent studies have suggested that colorectal cancer (CRC) screening is overused in older individuals with comorbid illness, a population of patients in whom the net benefit of screening is small. Efforts to curtail overuse may be limited by patients' reluctance to discontinue screening. It is not known if individualized risk/benefit estimates for screening can change attitudes towards discontinuation of screening. Methods: We developed a scenario-based survey that sought to elicit attitudes and preferences towards discontinuation of CRC screening. Potential survey participants were identified electronically using the endoscopic database at the VA Ann Arbor Healthcare System. This large integrated healthcare system provides screening colonoscopy services for a broad population of Veterans in southeast Michigan and northern Ohio. We identified participants who met the following criteria: (1) age ≥ 50; and, (2) prior complete, normal average-risk screening colonoscopy with adequate bowel preparation. Individuals were excluded if they had a personal/family history of colonic neoplasia or IBD. The survey was mailed to 1500 patients, with a second survey mailed to non-responders after 3 weeks. Results: 1089 of 1500 surveys (73%) were returned for analysis. Respondents were predominantly white (86%) and male (94%). The median age was 60-69 years. Most respondents (77%) thought it was reasonable to use age to determine when to start screening. On the other hand, less than half (48%) though it was reasonable to use age to decide when to stop screening, and 39% stated that they would never be comfortable stopping due to age. 45% reported that they would not be comfortable stopping screening even if they had serious health problems and their provider recommended it. Most respondents thought that risk calculators for life expectancy and colorectal cancer would be inaccurate. Initially, 90% of respondents reported that they would be uncomfortable with stopping screening before age 75. After being presented with hypothetical risk information suggesting that screening was of low benefit, fewer (50%) reported that they would be uncomfortable. In multivariable regression analysis, patient factors that predicted comfort with discontinuation of screening were: (1) male gender; (2) low perceived threat of CRC; (3) high perceived barriers to screening; (4) low perceived screening effectiveness; (5) high self-reported knowledge of CRC; (6) high trust in provider; and, (7) good self-reported health. Conclusions: Although many patients are reluctant to discontinue screening even if the benefit of screening is low, approximately half are receptive to discontinuation when presented with quantitative risk information.