A quick method to determine the content of phenolic compounds (CPT) in yerba mate (Ilex paraguariensis) was studied using near infrared spectroscopy (NIR), providing reductions of cost and analysis time and absence of laboratory effluents generation. A total of 111 samples of yerba mate from different regions of the State of Paraná (southeast, south central and metropolitan area) were analyzed according to the Folin-Ciocalteu method. A comparison with the proposed alternative approach, NIR spectroscopy, was performed in the spectral range of 4000-10000 cm-1. The average content of the CPT found by the conventional method was 84.82, 133.31 and 37.00 mg g-1, respectively, for the regions studied. Using the Principal Component Analysis (PCA), data sets of samples were successfully separated in three groups (regarding the different regions), where PC1 explained 99.96% of the variance, while PC2 and PC3 explained only 0.04% together. Models were developed using partial least squares (PLS) and some preprocessing strategies of the spectral data were evaluated. The best result was obtained by Multiplicative Scatter Correction (MSC) and first derivative with 6 latent variables. The best model showed a correlation coefficient (Rv) of 0.85 with a 12% prediction error regarding the samples of external validation (RMSEP). The results showed that the use of NIR combined with multivariate analysis can be applied as an efficient method for the analysis of CPT, allowing the sample classification in relation to its origin of planting, aiding the decision of referral and use of yerba mate in different final products.