This thesis work is focused on the evaluation of FT-NIR spectroscopy as an analytical technique for determining color and moisture content (%) of roasted and ground coffee samples of different varieties, geographical origins and roasting process procedures. Color and moisture are two of the most relevant parameters for assessing the quality of roasted coffee. From the obtained results, NIR spectroscopy has proven to be a reliable and robust analytical technique, faster and nondestructive than currently used reference methods, which can be implemented in the industrial process to simultaneously monitor multiple coffee quality parameters after each roast. NIR analysis of more than 400 samples made it possible, through the use of multivariate statistical analysis methods, to find the best pre-processing procedure for the obtained spectral data and to build robust calibration models for real-time and in situ determination of color and moisture content values.