Given the worldwide importance of soybean cultivars, it is essential to apply methodologies for the efficient monitoring of the physico-chemical parameters related to its quality. Nonetheless, the standard analytical methods used for it involves time-consuming techniques, usage of various equipment and reagents besides generating chemical residues. Considering this fact, the development of alternative methodologies can bring benefits to both industries and analysts, ideally offering relatively low cost, non-destructivity characteristics, minimally invasive being free of chemical reagents. In this way, this study aimed for the development of multivariate calibration models using Near Infrared Spectroscopy (NIRS) associated with chemometric methods based on Partial Least Squares (PLS) to predict acidity in ground soybean. For this, 300 soybean samples from Paraná State-Brazil, were evaluated. The model was mean-centered and cross-validated by contiguous block of 10 samples, using 12 latent variables. The model was validated by calculating its figures of merit, being accuracy based on Root Mean Square Error of Calibration (RMSEC)= 0.197, Root Mean Square Error of Prevision (RMSEP)= 0.169, fit for calibration (Rcal)= 0.859, Sensibility= 3.35x10-4, Analytical Sensibility= 18.01, Inverse of Analytical Sensibility= 0.055, Detection Limit= 0.183, Quantification Limit= 0.555, Relative Prediction Deviation for calibration (RPDcal)= 1.7661 and Relative Prediction Deviation for prediction (RPDval)= 1.8525. The results suggest that multivariate calibration models can be used to predict parameters in ground soybean in place of standard ones, providing satisfactory results. Taking into consideration the variability between these samples, the results suggest that the developed model can be used as an alternative methodology to determine acidity in ground soybean, once it would bring advantages such as decreasing the quantity of reagents used to this purpose, shorten the time for analysis and, consequently, increasing the production.