This study reports for the first time the use of (1)H NMR technique combined with chemometrics to study the metabolic profile of cocoa (Theobroma cacao L.) beans of different varieties, origin and fermentation levels. Results of PCA applied to cocoa bean (1)H NMR dataset showed that the main factor influencing the cocoa bean metabolic profile is the fermentation level. In fact well fermented brown beans form a group clearly separated from unfermented, slaty, and underfermented, violet, beans, independently of the variety or geographical origin. Considering only well fermented beans, the metabolic profile obtained by (1)H NMR permitted to discriminate between some classes of samples. The National cocoa of Ecuador, known as Arriba, showed the most peculiar characteristics, while the samples coming from the African region showed some similar traits. The dataset obtained, representative of all the classes of soluble compounds of cocoa, was therefore useful to characterise fermented cocoa beans as a function of their origin and fermentation level.