BACKGROUND: Authors have undertaken this study to see if the choice of anesthesia can directly or indirectly provide immunomodulation for cytokines, to determine the relationship of cytokines and hypothalamo-pituitary-adrenal axis in stomach cancer surgery patients, and also to see whether the amount of morphine administration and choice of analgesia can influence cytokine release, and possibly immunity. METHODS: Total 19 gastric cancer surgery patients were randomly assigned in double-blind fashion into two groups. Group-G (n=9) was provided with general anesthesia plus morphine intravenous patient controlled analgesia (IV-PCA), whereas group-GE (n=10) with preemptive epidural and general anesthesia plus continuous epidural analgesia for control of postoperative pain. At predetermined time interval, proinflammatory cytokines and stress hormones were evaluated with visual analog pain scale. Simultaneous assessments of operating and anesthesia time, total morphine doses, the time to recovery of gastrointestinal function and incidences of complications were also made. RESULTS: Demographic data, the durations of operation and anesthesia and recovery of gastrointestinal function were similar in both groups. Total morphine doses were approximately four times greater in group-G. Secretions of interleukin-1 beta , TNF and epinephrine were blocked by preemptive epidural anesthesia, meanwhile, interleukin-6 as well as ACTH and cortisol were not. After 24 hours after skin incision, the differences of cytokines, ACTH and cortisol between two groups were dissipated. In spite of these hormonal findings, visual analog pain scale could not disclose any differences. Incidences of complications were statistically insignificant except that of itching in group-GE. CONCLUSION: Preemptive epidural anesthesia and analgesia can partially block only some of cytokines and stress hormones, and these effects do not have clinically relevant long term influences. The amounts and means of morphine administered by continuous epidural analgesia block or IV-PCA demonstrated no evidence of immunosuppression at clinical dose range.