To evaluate the accuracy of MRI in predicting the composition of gallstones by Fourier transform infrared spectroscopy (FT-IR). Methods 100 patients with cholelithiasis who underwent pre-cholecystectomy were screened, and MRI of 50 patients without stones were taken as negative control. Preoperative MRI was performed, and postoperative cholecystolithiasis was analyzed by FT-IR. The correlation between FT-IR and MRI was analyzed, and McNemar test was used for comparison between matched sample count data groups. Kappa test, area AUC under ROC curve, sensitivity, specificity and accuracy were used to evaluate the accuracy of MRI stone composition prediction. Results Comparison of signal performance of each sequence of MRI with FT-IR: McNemar=4.00, P=0.135＞0.05, kappa=0.962＞0.7, P＜0.05, AUC (95% CI) was 0.980 (0.943-0.996), greater than 0.7, accuracy was 97.33%, sensitivity was 96.00%, specificity was 100.00%, and yoden index was 0.96. Comparison of signal values of each sequence of MRI with FT-IR: McNemar=0.20, P=0.655＞0.05, it can be considered that there is no difference in the detection rate between the two methods, kappa=0.953＞0.7, P＜0.05, AUC (95% CI) was 0.975 (0.935-0.994), greater than 0.7, accuracy was 96.67%, sensitivity was 95.00%, specificity was 100.00%, and yoden index was 0.95. In conclusion, there is no difference and good consistency between the two methods. Conclusion MRI can accurately predict the composition of gallstones, provide a new idea for clinical diagnosis and treatment of cholelithiasis, and provide a theoretical basis for the prevention and epidemiological study of cholelithiasis.