The aqueous solubility plays a key role in the biological processes that involved ionizable molecules like drugs, particularly, in the design and development of pharmaceutical formulations oriented to oral administration. Taking into account that experimental determination of aqueous solubility presents several limitations, like the elevated inverted time and the consumption of considerable quantities of sample, the use of structure-activity relationship (QSAR) studies has been suggested. These are intended to defined the function able to predict specific properties of a compound, using the information contained in their molecular descriptors. Therefore, it is possible to examine a great quantity of molecules in a minor time and with less resources. Then, the present work was oriented to obtain mathematical models for the prediction of aqueous solubility in pH range from 3.5 to 5.0. After processing a training serie of 602 compounds divided into five chemically different sub-series, five mathematical models were defined. In general, these models demonstrated a good prediction capacity after external testing. Four of them exhibited standard error of estimate inferior or close to the logarithmic unit. Also, the external prediction coefficients were superior to 0.8. Finally, the results obtained suggest these models for the design and development of new oral medicaments.