Background: Medically unexplained symptoms (MUS) are considered a common occurrence in medical settings, although definitions, methodologies and resulting prevalence rates for MUS vary widely between studies. Objectives: The objective of the present study was to characterize physicians' estimates of MUS, including clinically significant MUS, and to demonstrate in a single study how estimates vary based on the definition used. Methods: Two hundred and thirteen physicians completed an online questionnaire regarding the number of patients who present to their clinic with MUS. To reduce memory biases, participants reported on the number of patient seen in their most recent clinic day who met increasingly restrictive case definitions for MUS. Weekly estimates were also obtained. Results: The least restrictive definition yielded an estimate of 11%. When certainty criteria were added to the definition of MUS, the estimate decreased considerably to 4%. Approximately 3% of patients were estimated to have chronic MUS that affected their daily functioning or caused significant distress (i.e. psychologically significant MUS), and only half of these, 1.5%, were assigned a diagnosis of somatoform disorder or factitious disorder. The proportion of MUS cases accounted for by malingering was 18%. Conclusions: The present study documents significantly lower estimates of MUS than chart review studies. However, our results suggest that a significant proportion of the total number of patients who present with MUS have abnormal illness behaviour associated with significant impairment or distress. Despite physicians' recognizing significant distress and dysfunction in these cases, formal diagnoses of somatoform or factitious disorder are rarely assigned. © The Author 2010. Published by Oxford University Press. All rights reserved.