A set of 29 cannabinoid compounds with analgesic activity was studied by using the quantum chemical semi-empirical method AM1 and chemometric methods with the aim to correlate some calculated molecular properties (descriptors) of the compounds and their biological activity. The pattern recognition methods principal component analysis (PCA) and hierarchical clustering analysis (HCA) have been employed in order to investigate the possible relationships between the molecular descriptors and the analgesic activity. The PCA and HCA results showed that the descriptors charge density on substituent at position C 3 (R 3 ), charge on atom C 1 (Q 1 ), surface area (A), logarithm of the partition coefficient (logP) and molecular refractivity (MR) are responsible for the classification among the active, moderately active and inactive compounds. These results allow us to rationally propose new cannabinoid compounds as potential candidates for synthesis and biological evaluation regarding the analgesic activity.