The sonoextraction (SE) of aescin saponins and phenolic compounds from the inedible seed kernels of Aesculus hippocastanum was optimized using a central composite rotatable design coupled with response surface methodology, where the joint effects of ultrasonic power, sonication time, and ethanol proportion were investigated. Flavonol glycosides (≥90 % of the phenolic fraction), flavan-3-ols, phenolic acids, and aescin saponins were identified by HPLC-DAD-ESI/MS n , and the quantitative data was fitted to a quadratic model to predict the optimal SE conditions. After validating the models, the significant effect of the three factors was confirmed. The extraction of flavonols was maximized to 48 ± 2 mg/g extract by SE at 105.9 W for 4.1 min in 83.9 % ethanol, while 19.9 min sonication at 100.9 W in 95.8 % ethanol favored the recovery of 3.8 ± 0.1 mg/g extract of aescin saponins. A process for simultaneous SE of both classes of phytochemicals was also established. Overall, these SE processes proved to be time-saving and selective for the A. hippocastanum seed active constituents, which are cognized for their wide range of bioactivities and applications in the pharmaceutical, cosmetic, and food industries. • Horse chestnut seed kernel phytochemicals were characterized by HPLC-DAD-ESI/MS. • Flavonol glycosides, flavan-3-ols, phenolic acids and α- and β-aescin were identified. • Extraction was affected by sonication power and time and mainly by ethanol proportion. • Optimized sonoextraction processes were implemented for all classes of compounds. • Sonoextraction stood out as time-saving and selective for the seed’s active compounds.