An experimental model was used which includes intragastric instillation of 80 mM HCl and 0.6 ml bile/kg followed by intravenous infusion of live E coli in cats for up to three hours. This procedure regularly induces gastric mucosal ulcerations. Mucosal blood flow was measured by microspheres before, early, and late in sepsis. Total gastric blood flow was recorded electromagnetically. Mucosal regeneration capacity as reflected by the RNA/DNA ratio was measured. Misoprostol (a PGE1 analogue) was infused iv (5 micrograms/kg X h) or given locally in the stomach (10 micrograms/kg) before bacteriemia. Misoprostol did not influence the haemodynamic response to bacteria. The gastric mucosal damage was assessed either as an index representative for the entire corpus-fundus region or as the number of areas with intact surface epithelium within the series. Misoprostol iv protected the mucosa from ulceration compared with untreated septic controls while misoprostol intragastrically significantly reduced the number of damaged areas only. Topical misoprostol increased total gastric and mucosal blood flows early in sepsis compared to iv or no pretreatment while no difference was seen during late sepsis. The protective effect of misoprostol was thus not dependent on increased gastric mucosal blood flow. Nor was it mediated through effects on mucosal nucleic acid concentrations or ratio.