Objective. Repeat gastroscopy is not recommended for patients without alarm symptoms and with a normal earlier gastroscopy. However, there is little information available on the consequences of this recommendation. The objective of this study was to examine the role of earlier gastroscopy results in predicting the findings at repeat gastroscopy. Material and methods. Patients with previous gastroscopies presenting for a new gastroscopy during 2004–05 were included consecutively. A total of 293 patients who had undergone a gastroscopy a mean of 7.7 years (range 0.6–25.4 years) before the present gastroscopy were included in the study. The patients completed a questionnaire. The associations between the findings of the present gastroscopy and the findings of the previous gastroscopy and other patient characteristics were analysed by stepwise logistic regression. Results. Nine percent of the patients were positive for Helicobacter pylori infection. An abnormal macroscopic finding, defined as any erosion, ulcer or other macroscopic finding with the exception of hiatus hernia, at the repeat gastroscopy was significantly positively associated with: 1) an abnormal finding at a previous gastroscopy (OR 2.94, 95% CI 1.48–5.85), 2) obesity (body mass index, BMI >30) (OR 2.89, 95% CI 1.28–6.55), 3) the presence of alarm symptoms (OR 2.68, 95% CI 1.29–5.56), and negatively associated with 4) the use of proton pump inhibitors (OR 0.48, 95% CI 0.24–0.98). The findings were not associated with age. Conclusions. Abnormal earlier gastroscopy findings, obesity and the presence of alarm symptoms were the strongest indicators of abnormal findings at repeat gastroscopy. Our results support a restrained gastroscopy policy in patients with no alarm symptoms and a normal earlier gastroscopy.