Sweet taste is mediated via the Family C GPCRs dimer T1R2/T1R3. Most known sugars and sweeteners bind to the Venus Fly Trap extracellular domain of the T1R2 subunit. A possible approach for finding new sweet molecules is through structure-based virtual screening by docking to sugar-binding site. Since no experimental structure of human sweet taste receptors is available yet, we evaluated several homology models and docking protocols by their ability to differentiate between true positives and decoys. The best performing models were then used for prospective virtual screening, uncovering newly patented sweeteners. Next, the model was used for rationalizing structure-activity relationship of compounds derived from licorice, and to discriminate sweet from non-sweet licorice compounds. Non-sweet licorice-tasting compounds did not surpass the docking score of the known sweet compounds. Analysis of docked sweet saponins indicated that it is important to form hydrogen bonds with residues N44 or Y103. The saccharide moiety and the functional group at position C-30 were common among the sweet licorice compounds. The C-30 functional group formed hydrogen bonds with N44. While non-sweet compounds, had lower docking scores, and were oriented towards the lower lobe of the binding site. Finally, though typically a change in chirality strongly affects ligand–receptor interactions, we show that L- and D-glucose across a few concentrations are perceived as similarly sweet by humans, and that in cell-based functional assays, both enantiomers activate the human sweet taste receptor TAS1R2/TAS1R3. Docking suggested that glucose enantiomers can bind in either one of two subpockets of the VFT domain of TAS1R2, each overlapping with the predicted positions of monosaccharide units of sucrose. The compatability of each of the hydroxyl-rich enantiomer is enabled by multiple hydrogen-bond donors and acceptors in the binding subpockets.