Objective: To analysis the temporomandibular joint synovial fluid (SF) sample by using Fourier transform infrared spectroscopy (FTIR) method

so as to investigate its application value in diagnosis and differential diagnosis among temporomandibular joint diseases. Methods: A total of 115 temporomandibular joint SF samples were collected

with 15 cases in the normal group

32 cases of anterior disc displacement without renaturation (ADDWR)

41 cases of temporomandibular joint osteoarthritis (TMJOA)

and 27 cases of idiopathic condyle absorption (ICR). Dry films were made from SF samples and principal component analysis (PCA) was used to separate samples from groups. A partial least squares-discriminate analysis (PLS-DA) was used to establish predictive model and evaluate predictive ability. Results: Based on the PCA results

TMJOA and ICR was able to be significantly distinguished from ADDWR and normal SF samples. Components in ADDWR and normal SF samples were similar. In PC1 loading

the main differentiated absorption peaks sited around 3 386 cm-1

2 943 cm-1

2 912 cm-1

1 682 cm-1

1 537 cm-1and 1 468 cm-1. Through partial least squares-discriminant analysis

a satisfactory prediction result was obtained. The internal cross-validation accuracy was 90%

and the external validation classification accuracy was 97%. Conclusion: FTIR spectroscopy combined with appropriate stoichiometry has great potential for assisting the clinical diagnosis and differential diagnosis of temporomandibular joint disease.