OBJECTIVES: Prior web-based utility elicitation methods have not been validated against the gold standard: an interviewer administered survey. We describe the design and validation of a new web-based standard gamble interview, using the interviewer administered standard gamble as the reference standard. METHODS: We sought to elicit utilities for three health states: current health; carrier of a susceptibility gene for colon cancer (otherwise healthy); colorectal cancer. A health state classification system was built using focus group interviews from members of a population-based registry containing colon cancer patients, relatives of colon cancer patients, and controls. Focus group participants completed an interviewer administered standard gamble for a sample health state, and then reviewed a web-based standard gamble interview developed for the study. Refinements were made based on comments. Next, 75 new participants from the registry were recruited to complete both surveys. Participants were randomly assigned to start with the interview or the web version. They were also asked to access the web site from home in 3 weeks and retake the survey. RESULTS: Participant ages ranged from 20 to 79. Forty percent were male; 18 (24%) had colon cancer, 20 (27%) were relatives; 37 (49%) were controls. 72 (96%) and 68 (91%) completed the interviewer-administered and web-based gambles, respectively. Forty-six (61%) completed the follow-up web survey. Mean (std. error) utility scores for the interviewer administered gamble were as follows: current health 0.89 (0.01); gene carrier 0.88 (0.01); colon cancer 0.75 (0.02). Scores for the web-based gamble were as follows: current health 0.88 (0.01); gene carrier 0.90 (0.01); colon cancer 0.73 (0.02). Utilities were not significantly different between methods for all health states. Test-retest utilities from the web survey were also not significantly different. CONCLUSIONS: The interviewer administered and web-based utility methods showed similar results. Web-based utility elicitation appears to be feasible and accurate.