In the first 3 yr of an uncensored open access gastroscopy service in a County Hospital, 891 patients attended for first gastroscopy. The data on these patients is presented and compared with a randomly selected group who attended for gastroscopy in the yr prior to the establishment of the service having come to the normal Consultant clinics. In the open access group the gastroscopy examination was normal in 29 per cent (32 per cent comparator group), 31 per cent had major abnormalities (33 per cent comparator group) and 40 per cent had minor abnormalities (35 per cent comparator group). Delay time from referral to endoscopy was 37 days for open access patients (45 days comparator group). Only 6 per cent of open access patients were brought back to O.P.D. (47 per cent comparator group) and 72 per cent of open access patients returned directly to their family doctor (28 per cent comparator group). A comparison of the Clonmel findings with British centres reporting their results shows a broadly similar picture. It is concluded that almost 1,300 unnecessary clinic visits were avoided by the provision of the open access service, some reduction in delay time to gastroscopy was achieved, the family doctor maintained control of patient management in the great majority of patients, the pattern of referral was not inappropriate and compared very well with the comparator group. Over the 3 yr there was a large increase in the number of gastroscopies performed which caused resource difficulties. It is recommended that adequate planning of these requirements should be carried out before an open access service is started. At least 1 additional dedicated gastroscopy only endoscopy service per week would be required.