Malaria is one of the most deadly diseases, affecting million of people especially in developing countries. Because of the rapidly increasing threat worldwide of malaria epidemics multidrugs resistant to therapies, there is an urgent global need to discover new classes of antimalarial compounds. In an effort to overcome this problem, we have investigated the use of structure-based classification models for the 'rational' selection/identification or design/optimization of new lead antimalarials from virtual combinatorial data sets. In this sense, TOpological MOlecular COMputer Design strategy (TOMOCOMD approach) has been introduced in order to obtain two quantitative models for the discrimination of antimalarials. A collected data set containing 597 antimalarial compounds is presented as a helpful tool not only for theoretical chemist but for other researchers in this area. The validated models (including non-stochastic and stochastic indices) classify correctly more than 90% of compounds in both training and external prediction data sets. They showed high Matthews' correlation coefficients; 0.87 and 0.82 for training and 0.86 and 0.79 for test set. The TOMOCOMD-CARDD approach implemented in this work was successfully compared with two of the most useful models for antimalarials selection reported so far. Thus we expect that these two QSAR models can be used in the identification of previously un-known antimalarials compounds.