Tight glucose control (TGC) using a sliding scale based on intermittent blood glucose measurements occasionally can have a fatal outcome as a result of insulin-induced hypoglycemia. The present study was undertaken to examine whether the use of an artificial pancreas to achieve TGC would be possible in postoperative patients with sepsis. The retrospective study was carried out as an exploratory study, focusing on the possibility of precise evaluation of the significance of TGC as a beneficial intervention by serological monitoring of various mediators. TGC was accomplished using an artificial pancreas (STG-22; (Nikkiso, Tokyo, Japan). The patients were divided into two groups: the TGC group (6 patients with sepsis in whom the target blood glucose level set at <150 mg/dl was attempted using the artificial pancreas), and the glucose control (GC) group (6 patients with sepsis in whom glucose control was attempted using a sliding scale; target blood glucose level was set at 200 mg/dl or lower). The mean blood glucose level was 129.7 ± 9.7 mg/dl in the TGC group and 200.9 ± 14.7 mg/dl in the GC group (P < 0.01, ANOVA). No hypoglycemia associated with the artificial pancreas was seen in any of the patients. The serum levels of S100A12 and HMGB-1 tended to decrease, and those of sRAGE tended to increase, in the TGC group. Further data collection from a larger number of cases would be expected to allow a precise assessment of TGC as a potentially beneficial intervention in sepsis patients.