Alzheimer’s disease (AD) threatens elderly human health and still lacks effective treatment. Our previous work showed that LGZGD possessed a neuroprotective effect on the Aβ25–35-induced neurotoxicity in differentiated PC12 cells, indicating that LGZGD may be a potential drug for treatment of AD. However, its pharmacodynamic substances which show anti-inflammatory and anti-oxidant stress activities are still unrevealed. This research aims to reveal the pharmacodynamic substances of LGZGD on Aβ25–35-induced PC12 cell model of AD based on a spectrum-effect relationship study by using HPLC-FT-ICR-MS method and multivariate statistical analysis. Firstly, the chemical composition spectra of different combinations of LGZGD were recorded by HPLC-FT-ICR MS. Subsequently, Aβ25–35-induced PC12 cell model of AD was established and pharmacodynamic experiments were conducted to evaluate their anti-inflammatory and anti-oxidant activities, respectively. Finally, the potential pharmacodynamic substances were screened out through spectrum-effect relationship study accompanied by multivariate statistical analysis including bivariate correlation analysis (BCA), grey relational analysis (GRA), principal component analysis (PCA), partial least squares regression analysis (PLSR). As a result, a total of 96 chemical consistents in different combinations of LGZGD were discovered. Among them, 7 components such as isoglabrolide, licorice saponin E2, licorice saponin N2 and licoisoflavanone were directly linked with the anti-inflammatory effects, and 14 constituents such as tumulosic acid, polyporenic acid C, dehydrotumulosic acid, dehydropachymic acid, and pachymic acid were directly correlated with the anti-oxidative stress activities. In conclusion, we combined the HPLC-FT-ICR-MS spectra with pharmacodynamic indicators to develop the spectrum-effect relationships of LGZGD for the first time, and successfully revealed its potential pharmacodynamic substances in the treatment of AD from the anti-inflammatory and antioxidant pathways in the cell model.