Abstract. Quality of gambir is mainly determined by its catechin content. While trading in traditional market, catechin content of gambir is predicted by a treatment of expert experience which leads to subjective judgement. Meanwhile, quantitative analysis is carried out by using chemical method. However, this method is destructive, high cost and time consuming, thus not suitable for real time measurement. The objective of this study was to develop calibration model to predict catechin content in gambir non-destructively and rapidly using near infrared (NIR) spectroscopy with difference spectra pre-processing treatment and Partial Least Square (PLS) methods. Spectra pre-processing treatment method used were normalization between 0 and 1 (n01), first derivative Savitzky-Golay 9 points (dg1), second derivative Savitzky-Golay 9 points (dg2), combination between n01 and dg1, and combination between n01 and dg2. Determination of the optimum number of PLS factors was conducted based on value of consistency and Predicted Residual Sum Square Error (PRESS) of the validation set (V-set-PRESS). The best spectra pre-processing treatment method was n01 in combination with dg1. Evaluation of model demonstrated that the model could predict catechin content in gambir. The model had high value of correlation coefficient (r = 0.95), low values of SEC and SEP (3.56 and 3.27 respectively), and high value of ratio of prediction to deviation (RPD = 3.60). This study demonstrated that NIR spectroscopy had excellent potential as non destructive and rapid analysis to determine catechin content in gambir.