Herein we present results of QSAR studies of tyrosinase inhibitors employing one of the atom-based TOMOCOMD-CARDD (acronym of TOpological MOlecular COMputer Design-Computer Aided Rational Drug Design) descriptors, molecular quadratic indices, and Linear Discriminant Analysis (LDA) as pattern recognition method. In this way, a database of 246 organic chemicals, reported as tyrosinase inhibitors having great structural variability, was analyzed and presented as a helpful tool, not only for theoretical chemists but also for other researchers in this area. In total, 12 LDA-based QSAR models were obtained, the first six with the non-stochastic total and local quadratic indices and the six remaining models with the stochastic molecular descriptors. The best two models for the non-stochastic and stochastic molecular descriptors, showed an appropriate overall accuracy (92.68 and 89.10%, respectively) and a high Matthews correlation coefficient (C of 0.85 and of 0.84, correspondingly) when applied to the training set. External validation series were also used to validate the obtained models; the 91.67% (C=0.82) and 90.00% (C=0.78), were correctly classified, respectively. In order to show the possibilities of the present approach for the ligand-based virtual screening of tyrosinase inhibitors, the developed models were used afterwards in a simulation of a virtual search for tyrosinase inhibitors. For instance, more than 93% (93.33%) and 96% (96.66%) of the screened chemicals were correctly classified by the two best LDA-based OSAR models developed with non-stochastic and stochastic quadratic indices, respectively. Finally, the combination of the obtained models permitted the selection/identification of new diterpenoidal alkaloid leads as tyrosinase inhibitors. The found activity is supported by observed inhibitory effects on mushroom tyrosinase enzyme, even comparable with some reference tyrosinase inhibitors. These results support a role for TOMOCOMD-CARDD descriptors in the biosilico discovery of novel tyrosinase inhibitors from large databases of chemical structures (virtual or in silico), which may be used to prevent or treat pigmentation disorders.