Esophagitis occurs in patients with excessive acid and/or alkaline gastroesophageal reflux. This observation prompted us to develop a continuously perfused in vivo rabbit esophageal model to examine the potential for different endogenous injurious agents to cause H+ back diffusion and morphologic evidence of esophagitis. We found that HCl at physiologic pH values did not break the mucosal barrier to H+ back diffusion or cause esophagitis. Bile salts at physiologic concentrations in both an acid or alkaline perfusate broke the mucosal barrier and caused H+ back diffusion, but failed to cause a morphologic injury consistent with clinical reflux esophagitis. Instead, proteolytic enzymes, such as pepsin in an acid environment and trypsin in an alkaline environment, caused a severe hemorrhagic erosive esophagitis consistent with that seen clinically. We feel new therapeutic strategies for the treatment of reflux esophagitis should be directed at proteolytic enzymes rather than only HCl or bile salts. Finally, we showed sucralfate to be a mucosal protectant against the acid-pepsin injury.