Objective: To study the time course and prediction of responses to reassurance after gastroscopy showing no serious illness. Design: Selection of consecutive patients were assessed before gastroscopy, immediately after reassurance, and at follow up at 24 hours, 1 week, 1 month, and 1 year. Responses of subgroups of patients identified as high, medium, and low health anxiety by the health anxiety questionnaire were analysed. Setting: Endoscopy clinic in a general hospital. Intervention: Oral reassurance that there was "nothing seriously wrong." Subjects: One consultant physician and 60 patients aged 18-74 referred for gastroscopy. Main outcome measures: Physician's and patients' ratings of the extent of the reassurance and patients' ratings of their anxiety about their health and of their illness belief. Results: There was good agreement between the patients and the physician about whether reassurance had been given. Health anxiety and illness belief decreased markedly after reassurance. Patients with high health anxiety showed a significant resurgence in their worry and illness belief at 24 hours and 1 week, and these levels were maintained at 1 month and 1 year later. Patients with medium levels of health anxiety showed a reduction in worry and illness belief after reassurance, and this was generally maintained during follow up. Patients with low health anxiety maintained low levels of health worry and illness belief throughout the study. Partial correlation analyses showed that the levels of worry and illness belief after reassurance were predicted by the health anxiety questionnaire. This measure also had predictive value beyond that of a measure of general anxiety. Conclusions: Medical reassurance results in a reduction of worry about health and of illness belief, but this may be very short term. Measurable individual differences in health anxiety can be used to predict the response to reassurance.