Telomerase is a reverse transcriptase enzyme that activates in more than 85% of cancer cells and it is associated with the acquisition of a malignant phenotype. Some experimental strategies have been suggested in order to avoid the enzyme effect on unstopped telomere elongation. One of them, the stabilization of the G-quartet structure, has been widely studied. Nevertheless, no QSAR studies to predict this activity have been developed. In the present study a classification model was carried out to identify, through molecular descriptors with structural fragments and groups information, those acridinic derivatives with better inhibitory concentration on telomerase enzyme. A linear discriminant model was developed to classify a data set of 90 acridinic derivatives (48 more potent derivatives with IC(50) < 1 microM and 42 less potent with IC(50) > or = 1 microM). The final model fit the data with sensitivity of 87.50% and specificity of 82.85%, for a final accuracy of 85.33%. The predictive ability of the model was assessed by a prediction set (15 compounds of 90% and 82.29% of prediction accuracy); a tenfold full cross-validation procedure (removing 15 compounds in each cycle, 84.80% of good prediction) and the prediction of inhibitory concentration on telomerase enzyme for external data of 10 novel acridines (90% of good prediction). The results of this study suggest that the established model has a strong predictive ability and can be prospectively used in the molecular design and action mechanism analysis of this kind of compounds with anticancer activity.