Objective: To study application of partial least squares in fourier transforn infrared spectroscopy(FT-IR) spectrum analysis for Angelica sinensis at different growth stages. Method: Pretreatment method of orthogonal signal correction and wavelet compression(OSCW) was used to reject uncorrelated variables in the original spectra before partial least squares(PLS) analysis. the first two principal components were employed to cluster samples of A. sinensis at different growth stages. Result: All samples are properly classified into three categories according to their growth stage

the results of clustering was closely related to their growing period

which reflect the differences in relative content of main chemical composition among the samples from various growing period. Conclusion: The proposed method that established with orthogonal signal correction plus wavelet compression can decrease noise of FT-IR spectrum and help to improve the clustering performance.