[Objective]To establish improved method described in Chinese Pharmacopoeia which could be used for determining the content of total tannin in RADIX SANGUISORBAE from different resources. [Methods]Improved ultraviolet spectroscopy described in Chinese Pharmacopoeia was used to determine the content of total tannin and the results were clustered by SPSS19. 0. [Results]The linear regression equation was as follows: Y = 115. 07X + 0. 098 1,R = 0. 999 3(n = 6). When the concentration of gallic acid was 1. 007- 10. 070 mg /L,it showed a good linear relationship with the absorbance. The average recovery rate was 100. 12% and the RSD was 2. 74%(n = 6). It was of good repeatability and stability. The extraction ratio was high and the results were accurate. S13(Fagopyrum dibotrys(D. Don) Hara) was clustered into a class alone and its content of tannin was 2. 82% which was very low. In category 2,S19(Geranium strictipes R. Knuth) was clustered into a class alone and its content of tannin was 25. 63% which was high. The content of total tannin in S3 sample was 18. 15% which was relatively high and it was clustered into class 2. The content of tannin in RADIX SANGUISORBAE originated from Chuanzhusi,Songpan County,Sichuan was close to that of S3 and it was 18. 21%. And it was also clustered in the class. [Conclusions]The content of total tannin in samples from different sources differed significantly due to family,genus and species,collecting and distributing centre and degree of frying. After processing,the content of total tannin in medicinal materials decreased significantly. Species should be classified and quality control should be applied when RADIX SANGUISORBAE was used in clinic. By doing so,we could guarantee the safety and effectiveness of its clinical medication. And its clinical medication would be controllable. [Objective] To establish improved method described in Chinese Pharmacopoeia which could be used for determining the content of total tannin in RADIX SANGUISORBAE from different resources. [Methods] Improved ultraviolet spectroscopy described in Chinese Pharmacopoeia was used to determine the content of total tannin and the results were clustered by SPSS 19. 0. [Results] The linear regression equation was as follows: Y = 115. 07X + 0. 098 1, R = 0.999 3 (n = 6). When the concentration of gallic acid was 1 It was a good linear relationship with the absorbance. The average recovery rate was 100. 12% and the RSD was 2. 74% (n = 6). It was of good repeatability and stability The extraction ratio was high and the results were accurate. S13 (Fagopyrum dibotrys (D. Don) Hara) was clustered into a class alone and its content of tannin was 2.82% which was very low. In category 2, S19 ( Geranium strictipes R. Knuth) was clustered into a class alone and its content of tannin was 25. 63% which was high. The content of total tannin in S3 sample was 18. 15% which was relatively high and it was clustered into class 2. The content of tannin in RADIX SANGUISORBAE originated from Chuanzhusi, Songpan County, Sichuan was close to that of S3 and it was 18. 21%. And it was also clustered in the class. [Conclusions] The content of total tannin in samples from different sources differed significantly rather to family, genus and species, collecting and distributing center and degree of frying After processing, the content of total tannin in medicinal materials decreased significantly. Species should be classified and quality control should be applied when RADIX SANGUISORBAE was used in clinic. By doing so, we could guarantee the safety and effectiveness of its clinical education. And its clinical will not be able to controllable.