Abstract: OBJECTIVE To provide information for the reasonable clinical application of antibiotics, a monitoring method based on multivariate data analysis (MVA) was developed. METHODS The original defined daily doses (DDDs) of 81 antibiotics in 12 quarters from 2011 to 2013 were used to establish a principal component analysis (PCA) model. The score plot and distance to model X block (DModX) control plot combination with contribution plots were applied to monitor the application of antibiotics in different quarters. The quarter-to-quarter consistency was evaluated and the causes of abnormalities were further diagnosed. RESULTS There was good quarter consistency between the 4th quarter of 2011, 1st-4th quarters of 2012 and 2013. The 1st-3rd quarters of 2011 deviated from the control limit in the DModX control plot, which were caused by the higher DDDs of 3 antibiotics (cefoselis, fusidic acid and norvancomycin), which had special requests in use, 8 antibiotics (cefminox, piperacillin/sulbactam, amoxicillin/sulbactam, furbenicillin, cefprozil, ceftizoxime, isepamicin and ornidazole), which had restrictions in use, and 4 antibiotics (cloxacillin, cefalexin, cefadroxil and cefotaxime), which had no restriction in use. CONCLUSION This study demonstrates that it is effective to employ MVA to monitor the clinical application of antibiotics. This research provides a new scientific tool for the monitoring of antibiotics.