Volatile profiles of 63 black and 38 green teas from different countries were analysed with Proton Transfer Reaction-Time of Flight-Mass Spectrometry (PTR-ToF-MS) both for tea leaves and tea infusion. The headspace volatile fingerprints were collected and the tea classes and geographical origins were tracked with pattern recognition techniques. The high mass resolution achieved by ToF mass analyser provided determination of sum formula and tentative identifications of the mass peaks. The results provided successful separation of the black and green teas based on their headspace volatile emissions both from the dry tea leaves and their infusions. The volatile fingerprints were then used to build different classification models for discrimination of black and green teas according to their geographical origins. Two different cross validation methods were applied and their effectiveness for origin discrimination was discussed. The classification models showed a separation of black and green teas according to geographical origins the errors being mostly between neighbouring countries.