OBJECTIVE：To establish UPLC fin gerprint of 32 compatible herb pairs with Polygonum multiflorum as the core ， and to conduct multivariate statistical analysis. METHODS ：UPLC method was adopted. Using P. multiflorum and single decoction pieces of compatible herb as reference ，UPLC fingerprints of 32 compatible herb pairs of P. multiflorum were drawn. Common peaks were confirmed by relative retention time and UV absorption spectrum. Non-supervised PCA and supervised OPLS-DA were conducted by using SPSS 19.0 software and SIMCA 13.0 software. RESULTS ：There were totally 12 common peaks in UPLC fingerprints of 32 compatible herb pairs of P. multiflorum . The results of non-supervised PCA showed that the accumulative variance contribution rate of primary 6 principal components was 84.633％. The results of cluster analysis of PCA comprehensive score showed that single decoction piece of P. multiflorum ，compatible herb pairs of P. multiflorum with Lycium barbarum ，Rehmannia glutinosa，Paeonia lactiflora ，Codonopsis pilosula ，Eclipta prostrate ，Angelica sinensis ，Glycyrrhiza uralensis ，Astragalus membranaceus and Ophiopogon japonicus were clustered into one category ；others were clustered into one category. Results of supervised OPLS-DA analysis showed that eigen values of 4 principal components were 2.32，2.61，1.58 and 0.90，respectively. There were differences in the contents of 12 common components in the compatibility of P. multiflorum with tonic medicine and non-tonic medicine. The changes of the content of the components after compatibility with tonic medicine were similar. Common peak 7，4，6，3 were main reasons for the differences （variable importance projection value were all higher than 1）. CONCLUSIONS：Established fingerprint is simple in operation ，and can be combined with multivariate statistical analysis to evaluate the content changes of common components of 32 compatible herb pairs with P. multiflorum as the core.