BACKGROUND: Video-recorded consultations are widely used for research in general practice. Recently, video recordings have begun to be used for the purposes of general practitioner (GP) registrar assessment. It is unknown, however, whether consultations in which patients withhold consent for recording differ from those that are recorded. AIM: To compare clinical problems and demographic characteristics of adult patients who consent to the video recording of consultations with those who withhold consent. METHOD: This was prospective study of 538 adult patients consulting 42 GPs, based in practices throughout Leicestershire. Each patient attended a surgery session with one of the 42 GPs between April 1995 and March 1996. Clinical presentations and demographic characteristics of patients consenting and withholding consent to the video recording of their consultations were compared. GPs' perceptions of whether patients in these two groups were distressed/upset or embarrassed were also compared. RESULTS: A total of 85.9% (462/538) of adults consented to video recording, and 14.1% (76/538) withheld consent. Multiple logistic regression revealed that patients who presented with a mental health problem were more likely to withhold consent to recording (odds ratio 2.5, 95% confidence interval 1.4-4.6). Younger patients were also more likely to withhold consent to video recording. Additionally, where patients' consent was withheld, GPs perceived patients to be more distressed or embarrassed. CONCLUSION: Younger patients and those suffering from mental health problems are more likely than others to withhold consent to being video recorded for research purposes in general practice. The implications of this study for the assessment of registrar GPs using video-recorded consultations are discussed.