NR-CBZ was prepared from a ternary mixture in which Naringenin (NR), carbamazepine (CBZ), and NR-CBZ all exist. In order to control the quality of NR-CBZ, vibrational spectroscopy (ATR-FTIR and Raman spectroscopy) in conjunction with PLS and PCR has been successfully applied to the simultaneous quantitation of NR, CBZ, and NR-CBZ in ternary mixtures. Finally, from a comparison of the predictive performance and error analysis, it was found that Raman spectroscopy performs with higher accuracy at the whole spectral range. The PLS model with Savitzky-Golay filter algorithms at the whole spectral range resulted in a better result which predicted the concentration of Naringenin, carbamazepine, and NR-CBZ. Display omitted •The first quantitative method of naringenin-carbamazepine cocrystal by spectrums.•The PLS and PCR were applied to cocrystal’s quantitative, firstly.•Several pre-processing algorithms were evaluated systematically. Naringenin, commonly found in citrus fruits, is one of the active pharmaceutical ingredient (API) in naringenin-carbamazepine drug-drug cocrystal (NR-CBZ). In the preparation of cocrystal, quantitative determination of NR-CBZ is essential for the quality control. In this paper, ATR-FTIR and Raman spectroscopies combined with partial least squares (PLS) and principal component regression (PCR) were used to quantify NR-CBZ in the mixtures. To improve the accuracy of the prediction models, median, denoising, multiple scattering correction (MSC), first and second order derivatives, Savitzky-Golay filter and SNV were used and their performance was evaluated using prediction errors in combination with different spectral ranges. Raman spectra, PLS combined with Savitzky-Golay filter over the entire spectral range were found to determine the best content prediction result for NR-CBZ. The root mean square error of cross-validation (RMSECV), root mean square error of prediction (RMSEP) and squared correlation coefficient (R2) of the model were 0.101, 0.132 and 0.870, respectively.