While meteorologic conditions are thought to be related to the incidence of certain cardiac and cerebrovascular events, very little information is available concerning the onset of variceal bleeding. We undertook this study to determine whether there is any correlation between esophageal variceal hemorrhage and several meteorologic and astronomic parameters.Sixty-seven episodes of bleeding from ruptured esophageal varices observed in 52 patients from southern Tunisia during a 7-Year period were studied retrospectively. All patients underwent endoscopic examination within 24 hours of admission. The date of the event (variceal rupture) was considered--if no stigmates of active bleeding were found at endoscopy--as the day of occurrence of the hematemesis and the day preceding melena. Meteorologic readings (11 parameters) plus moon-cycle, circadian and seasonal distribution of the 67 days of variceal bleeding in comparison with a control period of days randomly selected among 134 days without rupture.No relationship was found between variceal bleeding and mean atmospheric pressure, daily hours of sunshine, nebulosity, direction and velocity of wind and mean humidity. On the contrary, a significant correlation was observed with the mean temperature (18 degrees C vs 21.16 degrees C; p=0003), rainfall (p<0.01) and stormy weather (p=0.008), the latter being the only parameter retained as an independent factor at multivariate analysis: OR=13.37 (95% CI=1.5-118.5). Furthermore, a negative correlation was found with full moon but only at univariate analysis (p=0.04). The seasonal distribution of variceal bleeding episodes showed the highest percentage during winter with a significant variation at multivariate analysis: OR=3.2 (95% IC=1.6-6.54). Occurrence of variceal bleeding also showed a circadian variation with the higher prevalence night between 6 and 0 PM (p<0.001).