The seasonal fluctuations of acute upper gastrointestinal bleeding treated from 1991 to 1996 in Patras, Greece, were analyzed retrospectively. During that period, 1992 patients with acute upper gastrointestinal bleeding were admitted to our hospital. After patients who were not residents of the region served by our hospital were excluded, the remaining 1879 cases were reviewed. We observed seasonal fluctuation with low prevalence in winter and an increase in spring and autumn with two peaks in April and October (p < 0.00001). The seasonal prevalence parallels that of duodenal ulcer bleeding, which follows a similar fluctuation (p < 0.00001). Bleeding due to gastric ulcers or other causes presented no periodicity. Seasonal fluctuation, both in total numbers of upper gastrointestinal bleeding and in duodenal ulcer bleeding, was statistically significant only in patients not receiving nonsteroidal anti-inflammatory drugs (p < 0.00001). We conclude that upper gastrointestinal bleeding shows a seasonal fluctuation parallel to duodenal ulcer bleeding and is not related to nonsteroidal anti-inflammatory drugs. The seasonal pattern supports the traditional view of duodenal ulcer exacerbations.