Kaixin San (KXS) is a classical prescription for the treatment of depression. However, the mechanism is not clear. In this study, serum pharmacochemistry, mediated by the UHPLC-Orbitrap Exploris 480 mass spectrometer, was used to identify compounds derived from the KXS-medicated serum. These components were used to construct a compound-target network for depression using a network pharmacology approach to predict potential biological targets of KXS. Subsequently, we established a mouse model of CUMS-induced depression and observed the antidepressant effect of KXS. The signalling pathways predicted by the network pharmacology were further validated in animal experiments. The results showed that 36 compounds were identified from the KXS-medicated serum. Based on this, 984 genes related to the compounds and 4966 genes related to depression were identified using network pharmacology. Critically, KEGG analysis identified the PI3K/Akt and NF-κB signalling pathways as the main pathways through which KXS exerts its antidepressant effect. KXS significantly alleviated depression-like behaviour and hippocampal histopathological changes in a mouse model of depression. Compared with the model group, the treatment of KXS significantly reduced the expression of protein targets in the PI3K/Akt/NF-κB signalling pathway. All these studies effectively corroborated the predicted results, confirming the feasibility of this integrated strategy.