Response rates for telephone surveys of the general population—and even more so for those targeting private-practice healthcare professionals—have been declining sharply for nearly a decade. To minimize refusals to participate, INPES implemented a series of measures as part of the "2009 General Practitioners' Health Barometer" survey, including the option of an alternative web-based survey. The sample of private practitioners was randomly selected from a database regularly updated by medical representatives and supplemented with various data, such as medical journal distribution and prescription orders (CEGEDIM®). This sample was contacted by telephone between November 2008 and January 2009. Any physician who declined the telephone interview or was unable to keep the scheduled appointment to complete the survey was contacted again; an online questionnaire, structured identically to the telephone questionnaire (question wording, filters, on-screen instructions for interviewers, etc.), was then offered to them. Thus, 1,836 general practitioners were contacted again to be offered this questionnaire. Of these, 129 ultimately agreed to respond by phone, 532 agreed to respond online (by providing their email address); in the end, 185 completed the web questionnaire, representing nearly 9% of the final sample (n=2,083) and 10% of those who initially declined. The overall response rate thus increased from 48.0% to 57.1%. In addition to describing the implementation of this web-based survey, the presentation will address the specific characteristics of the supplementary sample, particularly its structure by gender, age, practice location, and other variables characteristic of professional practice. This sample will be compared not only to the telephone sample but also to the structure of the population of private general practitioners available from other sources (SNIR, SNIRAM, ADELI). Furthermore, the difficulties associated with the coexistence of two data collection methods—particularly regarding the validity of responses—will be discussed. Ultimately, the aim will be to assess the extent to which different results would have been observed if this complementary web-based study had not been conducted. [author’s abstract]