Abstract: Objectives:The aim of this study was to screen the key components in synthetic medium that affect the efficient expression of xylanase in recombinant Pichia pastoris. Methods:Firstly,the best original defined medium was selected by employing single factor method and t-test. Secondly,the key components were screened via establishing the multivariate linear regression models between medium components and responses with Plackett-Burman design and stepwise regression. Finally,the concentration of non-critical medium components were determined with correlation coefficient analysis. Results:Glycerophosphate,calcium sulfate,PTM1 and ammonium sulphate were screened as key components. Furthermore,it was found that the enzyme activity and specific enzyme activity of xylanase reached maximum values of 2298.4 U/mL and 9926.3 U/mg in Plackett-Burman design,respectively(i.e. when glycerophosphate 20 g/L,(NH₄)₂SO₄ 2.0^g/L,CaSO₄·2H₂O 2.0 g/L,K₂SO₄ 20.0 g/L,MgSO₄·7H₂O 6.0 g/L,PTM1 8.0 mL/L,Methanol 10.0 mL/L,Tween 80 2.0 g/L),which were 3.055 and 3.889 times as many as before optimization,respectively. Conclusion:Plackett-Burman design not only significantly enhance the optimization objectives,but also screen the key components,and thus lay the foundation for further optimization.