OBJECTIVE--To investigate factors influencing a general practitioner's decision to do a rectal examination in patients with anorectal or urinary symptoms. DESIGN--Postal questionnaire survey. SETTING--General practices in inner London and Devon. SUBJECTS--859 General practitioners, 609 (71%) of whom returned the questionnaire. MAIN OUTCOME MEASURES--Number of rectal examinations done each month; the indication score, derived from answers to a question asking whether the respondent would do a rectal examination for various symptoms; and the confidence score, which indicated the respondent's confidence in the diagnosis made on rectal examination. RESULTS--279 General practitioners did five or fewer rectal examinations each month and 96 did more than 10 each month. Factors significantly associated with doing fewer rectal examinations were a small partnership and being a female general practitioner, and the expectation that the examination would be repeated. Lack of time in the surgery, and a waiting time of less than two weeks for an urgent outpatient appointment were also important. General practitioners were deterred from doing rectal examinations by reluctance of the patient (278), the expectation that the examination would be repeated (141), and lack of time (123) or a chaperone (39). Confidence in diagnosis was significantly associated with doing more rectal examinations, the perception of having been well taught to do a rectal examination at medical school, and being a male general practitioner. CONCLUSIONS--Factors other than clinical judgment influence the frequency of rectal examination in general practice. Rectal examination may become commoner with the trend towards larger group practices and if diagnostic confidence is increased and greater emphasis put on rectal examination in undergraduate and postgraduate teaching.