An experimental model was used in mice in which septicemia develops following invasion of the animals' own intestinal flora after cecal ligation and puncture. Pretreatment with 1 microgram of endotoxin administered 24 hours before surgery significantly reduced the rate of lethality. Bacteria were counted and differentiated in cardiac blood at various times throughout a 48-hour period after induction of septicemia in mice, with and without pretreatment. Endotoxin levels and plasma-related interference of the Limulus-amebocyte-lysate (LAL)-endotoxin reaction also were determined as were hematologic and metabolic parameters. A shift from mixed gram-positive and gram-negative to predominantly gram-negative bacteria occurred in both groups. In pretreated mice, a reduction in aerobic bacterial counts and reduced hyperglycemia were seen in the initial phase; and a decrease in anaerobic and aerobic bacteria and in endotoxin levels were observed at the end of the experiment. This appears to be related to endotoxin-induced increased resistance against the consequences of intraabdominal sepsis. These studies also indicate that the measured amount of circulating endotoxin does not necessarily correlate to the severeness of infection. Individual differences in plasma-related interference with the LAL-endotoxin reaction also emphasize the requirement for sample-internal standardization in order to reliably quantify endotoxin in plasma.