The degree of lipophilicity of a drug, defined by its partition coefficient, is a key parameter to understand and analyze the activity of said compound in the body. In the case of drugs active on the central nervous system, the value of the partition coefficient or log P indicates their ability to cross the blood-brain barrier and carry out their pharmacological action. However, the experimental determination of this property is accompanied by several drawbacks, since the methods used for this purpose are usually time-consuming, resource-intensive, and expensive equipment, in addition to being prone to measurement errors. To solve these problems, computational molecular modeling methods are used. Among them are the QSPR studies, with which quantitative relationships are established between the structural characteristics of the analyzed compounds and the property of interest. In the present work, a predictive model was built based on the partition coefficient as a property of interest. For this, a training series of 286 active drugs on the central nervous system was built, obtaining their structures, simplified representations and experimental values of the partition coefficient using the ACDLabs software. The MODESLAB program was used to calculate a total of 91 molecular descriptors and based on these, the BuildQSAR software was used to obtain the predictive model