Internet surveys of the general population have been growing in use over the last few years because they are less expensive, faster, and greener than most other survey modes. However, they have been used less frequently to survey elites, and little is known about the representativeness of their results. This article helps fill that void by using an experiment where half of a sample of state legislators are given an Internet survey and half a mail survey, and then the response rate, representativeness of respondents, and similarity of responses are compared. The results indicate that mail and Internet surveys generate fairly representative samples, albeit with few biases, and similar responses; however, mail surveys generate a significantly higher response rate than Internet surveys. The implications of these findings are discussed.