Several molecular properties are calculated for a set of 26 cannabinoid compounds with the goal of connecting the psychoactivity of the compounds with an appropriate set of calculated properties. For this purpose we used quantum chemical (the AM1 semi-empirical method) and chemometric methods. The AM1 method was employed to calculate the set of quantum chemical molecular properties and the chemometric methods were employed with the aim of selecting the most relevant properties to be correlated with psychoactivity. The chemometric methods used were Principal Component Analysis (PCA), Hierarchical Cluster Analysis (HCA) and the K-Nearest Neighbor (KNN) method. The chemometric analysis showed that an electronic property (energy of LUMO), a hydrophobic property (log P), a steric property (volume of the substituent at the C4 position) and a topological property (Lovasz-Pelikan index) were the most important variables for the separation between the psychoactive and psychoinactive compounds. In order to validate our PCA, HCA and KNN results, eight new cannabinoid compounds (with known psychoactivity) were used in a prediction study and were classified correctly by the methods used in this work, indicating that our PCA, HCA and KNN models are able to predict reliable psychoactivity of cannabinoid compounds.