Purpose: To evaluate the optimal dose of dexmedetomidine for perioperative blood glucose regulation in non-diabetic patients with gastrointestinal malignant tumor. Methods: One hundred patients were randomly divided into four groups: control group (group C), dexmedetomidine 1 mu g/kg + 0.25 mcg/kg/h (group D-1); + 0.5 mcg/kg/h (group D-2); and + 1 mcg/kg/h (group D-3). Blood glucose concentrations were measured before dexmedetomidine infusion (T-1), 1 h after surgery beginning (T-2), at the end of surgery (T-3), and 1 h in PACU (T-4). Duration of surgery, extubation time, anesthetics doses, adverse reactions, postoperative pulmonary infection, total peritoneal drainage 2 days after surgery and hospital stay were recorded. Results: Compared with T-1, blood glucose concentrations were higher at T-4 in group C and at T2-4 in groups D-1, D-2, and D-3 (p < 0.01). Compared with group C, blood glucose concentrations were higher at T-2 and T-3 in groups D-2 and D-3 (p < 0.05), but significantly lower at T-4 in groups D-1, D-2, and D-3 (p < 0.01). Propofol and remifentanil consumption in groups D-1, D-2, and D-3 decreased significantly compared with group C (p < 0.01). In group D-3, doses of ephedrine (p < 0.05) and atropine (p < 0.01) were higher, and extubation time was prolonged (p < 0.01) compared with the other groups. The incidence of bradycardia was higher in group D-3 than that in group C (p < 0.05). Conclusions: Dexmedetomidine loading dose of 1 mcg/kg followed by maintenance with 0.25 mcg/kg/h can regulate perioperative blood glucose well in non-diabetic patients undergoing gastrointestinal malignant tumor resection and reduce doses of anesthetics without extending extubation time.