Abstract: Conductor-like screening model for real solvents(COSMO-RS)model was employed to screen the suitable natural deep eutectic solvents(NADESs)for flavonoids extraction from Ginkgo biloba leaves. On the basis of the result from single factor experiment,Box-Behnken design from response surface methodology was used for process optimization with flavonoids yield as the main indicator. It was found that choline chloride and glycerol with molar ratio of 1:2 was the better NADESs for extraction. When using this NADESs with solid/liquid ratio of 1:30.7 (g/mL),at 73.2℃ for 4.1 h,the highest flavonoids yield of 5.70%was achieved from the prediction of the model. The result from experiment under such conditions was 5.68%,which fitted well with the prediction. Through combination of computational simulation and experiment,this study provided a reference for screening suitable solvents to extract bioactive products from natural resources,as well as further optimization of extraction process.