<p>Fast alternative techniques for assessing the chemical profile of cocoa beans in a single analysis and without the use of chemical reagents are of great interest for quality control, routine analysis, and as a means of indicating geographical origin. Mid-infrared (MIR) and near-infrared (NIR) spectroscopy, combined with multivariate statistical analysis, have proven effective in discriminating the origin of cocoa beans. This study aimed to evaluate the physicochemical composition and NIR/ MIR spectra, in combination with multivariate analysis, as a contribution to the Protected Geographical Indication (PGI) of cocoa beans produced in the state of Bahia, Brazil. The results obtained from the composition data showed a centesimal similarity between the samples and the difficulty in distinguishing them based on their cultivation region. However, it is possible to observe the formation of patterns according to the region using the MIR data associated with Principal Component Analysis (PCA), and that the use of MIR combined with Linear Discriminant Analysis (LDA) revealed of the best discrimination rates and classification of samples, demonstrating the potential of this technique for fast and simple classification of cocoa beans.</p> Graphical abstract <p></p>