OBJECTIVE： To observe clinical efficacy and safety of off-label use of tigecycline combined with cefoperazone and sulbactam in the treatment of pneumonia caused by multidrug-resistant Acinetobacter baumannii. METHODS： In retrospective study， 57 patients with pneumonia caused by multidrug-resistant A. baumannii were collected as research subjects from ICU of our hospital during Apr. 2015-Apr. 2016. They were divided into routine drug use group （28 cases） and off-label drug use group （29 cases）. Routine drug use group was given routine dose of tigecycline （initial dose of 100 mg， and then 50 mg each time， q12 h） combined with cefoperazone and sulbactam （4 g/d by sulbactam， 3 separate doses）； off-label drug use group was tigecycline with initial dose of 100 mg， and then 75 mg each time， q12 h， combined with cefoperazone 6 g/d by sulbactam， 3 separate doses according to Expert Comsensus on Antibiotic Use. Clinical efficacy， the rate of bacterial clearance， 28 d mortality， the occurrence of ADR， weaning success rate， ICU hospital stay and total length of stay were compared between 2 groups. RESULTS： After treatment， response rate， the rate of bacterial clearance and weaning success rate in off-label drug use group were significantly higher than routine drug use group （89.7％ vs. 67.9％， 69.0％vs. 46.4％， 75.9％vs. 53.6％）； 28 d mortality of off-label drug use group was significantly lower than that of routine drug use group （6.9％ vs. 17.9％）； ICU hospital stay and total length of stay were all significantly shorter than routine drug use group， with statistical significance （P＜0.05）. The incidence of ADR in off-label drug use group was similar to routine drug use group， without statistical significance （P＞0.05）. CONCLUSIONS： For pneumonia caused by multidrug-resistant A. baumannii， off-label use of tigecycline combined with cefoperazone and sulbactam is an effective treatment method because of significant therapeutic efficacy， effective bacteria removal， high weaning success rate， low 28 d mortality， significant decrease of ICU hospital stay and total length of stay， no change in the incidence of ADR.