The aims of the present study were to evaluate the performance of the main methods proposed for the comparison of percentage dissolved versus time curves and to recommend a more biorelevant combined approach for the comparison of dissolution profiles of multisource drug products. In vitro dissolution tests of four brands of oxcarbazepine (OxCBZ) tablets were performed, and the resulting profiles were compared by model-independent, model-dependent, and ANOVA-based statistical methods. After a careful analysis of the results, some methods were chosen and applied to the comparison of dissolution profiles of four brands of carbamazepine (CBZ) tablets and two brands of phenytoin (PHT) capsules. Finally, these in vitro results were qualitatively correlated with the corresponding in vivo results previously obtained with the same CBZ and PHT products assayed in healthy volunteers. The analysis of the dissolution data obtained with OxCBZ tablets allowed discarding the ANOVA-based statistical methods since in all cases they were over-discriminating from a biopharmaceutical point of view. The remaining comparison methods were applied to in vitro profiles of CBZ and PHT products and the results correlated with in vivo data. The most suitable methods for the biopharmaceutical comparison of in vitro dissolution profiles were the model-independent ones, and among them, the best correlations were the f2 similarity factor along with a measure of the dissolution extent (e.g., area under the curve). This combined approach gives a robust and informative result with the most biopharmaceutical relevance.