The Membership of the Royal College of General Practitioners (MRCGP) ‘Simulated Surgery’ examination, taken routinely by 90–140 candidates at each diet of the MRCGP examination to assess their consultation skills, comprises a controlled set of ten-minute consultations with trained role players acting as simulated patients (SPs), the performance on each being adjudged by an examiner. This paper reports on the quality assurance feedback data from candidates and role players in the December 2004 examination. The Simulated Surgery was overwhelmingly perceived by candidates as a fair, if challenging, assessment affording realistic and representative general practice experience. Some overseas candidates perceived difficulties with the approach, especially those who had undertaken their general practice vocational training abroad. Few role players experienced difficulty in ‘coming out of role’ they varied in their ability to detach emotionally from their role and in how stressful they perceived the experience; only one out of 24 role players (4%) found coming out of role difficult. Almost all felt that they understood the role of the general practitioner (GP) better and were more able to judge the latter's quality accurately as a result of their role-playing, but they felt that the quality of GPs' skills varied more than they had previously thought; and none was less likely to consult their GP or had determined to change their GP as a result of their experiences in the examination. Continuous role-playing is possible for three days in such an assessment: we believe that it may be the limit, but suggest that this matter should provide a focus for future work. [ABSTRACT FROM AUTHOR]