As telephone surveys have increased both in difficulty and cost, more research is migrating to the lower cost alternative of the Web. Yet the influence of mode, especially on health-related questions, remains unclear. This paper reports on an experimental comparison of web and telephone in a study of employee attitudes toward health plan alternatives. Respondents were randomly assigned to either telephone or Web as their primary mode. The mixed-mode study achieved an overall response rate of 84%. Telephone respondents exhibited social desirability effects for several health-related questions, including overall health status and reports of healthy behaviors. The use of multi-item indices seemed to mitigate this effect. The study also found higher item nonresponse on the web, but little evidence of modal differences in satisficing behaviors.