A method for validating discrimination and classification of honey samples performing GC/MS fingerprinting of headspace volatile compounds has been created. A number of varied honey samples from different plants and geographical regions around Greece were tested using orthogonal partial least squares-discriminant analysis (OPLS-DA, soft independent modelling of class analogy (SIMCA) and OPLS-hierarchical cluster analysis (OPLS-HCA). The results revealed an excellent separation between the honey’s tested depending on their botanical origin, and the misclassification was as low as 1.3%. Fragments which determined the separations were assigned to phenolic, terpenoid, and aliphatic compounds that are present in the headspace of unifloral honey. Furthermore, a variable classification for citrus and thyme honey’s relating to geographical origins could also be achieved. Thus results suggest that the developed methodology is both robust and reliable.