OBJECTIVE：To establish HPLC fingerprint of different products of suet oil-baked Epimedium brevicornum ，and to screen the optimal baking technology. METHODS ：HPLC method was adopted. Using icariin as reference ，HPLC fingerprints of 22 batches of samples were drawn. The similarity was evaluated by using Similarity Evaluation System of TCM Chromatographic Fingerprint（2012 edition），and common peaks were confirmed. HCA ，PCA and OPLS-DA analysis were performed by SIMCA 14.1 statistical software. Taking variable importance in the project ＞1 as criteria ，biomarkers affecting the quality difference of suet oil-baked E. brevicornum were screened ；using mass marker as index ，the baking technology was screened by baking technology. RESULTS：There were 18 common peaks in 22 batches of samples. The similarities were between 0.831 and 0.991. Totally 7 common peaks were identified ，i.e. epimedin A ，epimedin B ，epimedin C ，icariin，sagittatoside A ，sagittatoside B ，baohuoside Ⅰ. The 22 batches of samples were clustered into two categories ，S19-S22 were clustered into category Ⅰ and S 1-S18 were clustered into category Ⅱ. The category Ⅱ was sub-clustered into category Ⅱa（S15-S18），category Ⅱb（S10-S14），category Ⅱc （S1-S9）；the result of PCA analysis was consistent with above results. OPLS-DA showed that the biomarkers affecting the quality difference were icariin ，sagittatoside B and baohuoside Ⅰ. The results of kinetic studies showed that the content of icariin when baked at 180 ℃ for 25 min or 190 ℃ for 20 min，that of baohuoside Ⅰ when baked at 180 ℃ for 30 min or 190 ℃ for 15 min and that of epimedin B when baked at 210 ℃ for 18 min were the highest ；according to above results ，the optimal baking technology was baking at 180 ℃ for 25-30 min or 190 ℃ for 15-20 min. CONCLUSIONS ：Established fingerprint is stable ， reliable and reproducible. The multivariate statistical analysis can be used for the changes of chemical components in E. brevicornum under different baking condition and preliminary selection of baking technology.