Chocolate and cocoa-based products are among the goods with higher added value. A current trend of the cocoa market is to offer to the consumers high quality cocoa products, namely mono-origin cocoa. However, a reliable analytical method able to trace the geographical origin of cocoa is lacking. In this work we tested the capability of HR MAS H NMR combined with chemometrics to assess the geographical origins of 60 fermented and dried cocoa beans of 23 different cocoa producing countries from the three major crop-growing areas (Africa, Central/South America, Asia/Oceania). Metabolic profiling was determined by HR MAS H NMR directly on cocoa powder after the method optimization. The same samples were also subjected to extraction and analysis with HR H NMR. HR MAS H NMR, as H NMR analysis, allowed the simultaneous detection of amino acids, polyalcohols, organic acids, sugars, methylxanthines, catechins. Moreover, HR MAS allows the detection of lipids, not present in the aqueous extract utilized for H NMR. The data set obtained is therefore representative of all classes of cocoa compounds. Untargeted HR MAS H NMR and H NMR datasets were utilized as fingerprint of the samples and elaborated with multivariate statistical methods. A targeted quantitative approach of selected metabolites was possible only with HR H NMR data, because HR MAS H NMR does not give reliable quantitative results. All the approaches adopted showed a discrimination of the cocoa origins. HR MAS presents the advantages to obtain a very rapid picture of the samples, comprising both lipophilic and hydrophilic components, avoiding any sample manipulation.