This study compared the use of mail and Internet surveys of sentinel networks of physicians with traditional random sample mail surveys for three national vaccine policy surveys. Three nationally representative sentinel networks of physicians were established (pediatricians, n = 427; general internists, n = 438; and family physicians, n = 433). Surveys of the sentinel networks were compared with simultaneous surveys conducted with random samples of the American Medical Association (AMA) Physician Masterfile. Response rates were 74% to 78% for sentinel surveys and 29% to 43% for traditional random sample surveys. Respondents to the two methods were generally comparable in demographic characteristics. While there were some differences in responses to survey topic questions, none of the differences were likely to affect policy decisions. Sentinel networks represent the opinions and experiences of physicians in a manner equivalent to traditional mail surveys and may provide a more efficient approach to conducting physician surveys.