Within a few hours after one injection of fresh human serum by the intraperitoneal route only, mice developed pancreatic acinar cell necrosis and inflammation, fat necrosis, elevated serum amylase and a shocklike state. The extent of these lesions and mortalities were roughly dose dependent and were not noticeably modified by either different fasting cycles or pilocarpine. Acinar cell changes and necrosis usually developed first in subserosal acini. The earliest ultrastructural change detected was nonspecific swelling of cytoplasmic compartments which was reversible but also preceded the cytoplasmic degradation that developed in cells undergoing necrosis. Notably, zymogen granule dissolution neither preceded nor accompanied this swelling, but developed pari passu with cell degradation. Occasionally, intact granules were found in necrotic cells. Serum was cytotoxic for isolated acinar cells in vitro, even in the presence of soybean trypsin inhibitor. These results (1) indicate that the injury mechanism in vivo is directly initiated through contact of serum with acinar cell surfaces and is independent of zymogen secretions and trypsin activation, and (2) suggest that a rapid disturbance in cell membrane permeability results, the magnitude of which being the primary determinant of cell death. Pancreatic toxicity of human serum was abolished by aging, heating, ethylenediaminetetraacetic acid, heparin, zymosan, cobra venom factor, and absorptions with mouse red blood cells, against which fresh, unabsorbed serum was hemolytic. Pancreatic toxicity in vitro and, to a much lesser extent, in vivo was reconstituted by combining the red blood cell-absorbed serum with either heated serum, or with IgM-enriched, but not IgG serum fractions. Fresh cord serum was virtually nontoxic and could substitute for absorbed serum in such reconstitutions. These results indicate that the injury mechanism involves at least two serum components. By both circumstance and analogy, other results and a review of other examples of foreign sera toxicity suggest that they are components of a complement-dependent, cytotoxic heterophile antibody system. The relevance of this odd phenomenon is that it offers a simple model of acute pancreatitis, contributes to the debunking of traditional notions of the pivotal role of zymogens in the initiation of acute pancreatitis, and hints at a potential pathogenetic connection between pancreatitis and products of immune or related reactions.