To carry out accurate classification of four kinds of compound Danshen medicines and to propose the concept of accurate theoretical estimation of Chinese medicines efficiency.Methods:Characteristic sequence of each sample was determined relying on the dual index grade sequence individualized pattern recognition method,and the accurate identification of samples with similar composition of medicines was performed.Then the ratio of samples in correlation sequence of one samples characteristic sequence to the total samples，which belonged to the same class of the samples in the correlation sequence,was defined as the correlation degree parameter,in order to investigate theoretical approach for estimating efficiency of Chinese medicines with close composition.Results:The IR fingerprint spectra of 23 samples components,belonging to four kinds of similar compound formulae containing Salvia miltiorrhiza,extracted with chloroform were measured,and were analyzed relying on dual index grade sequence individualized pattern recognition method,and then their correlation degree analysis were carried out.As a result,these samples were accurately identified with the correct ratio is 96%.The correlation degrees among the four kinds of compound formulae were calculated also,and it was found that for the two pairs of compound formula among the four kinds,their efficiency was directly proportional to their correlation degree.The results proved that dual index grade sequence individualized pattern recognition theory was suitable to accurately identify many kinds of traditional Chinese compound formulae composed of very similar composition simultaneously.Conclusion:The correlation degree may be a good parameter for theoretically evaluating the quality and medical efficiency of traditional Chinese compound formulae.