Recently, scientists haveproven the green teatherapeutic effectsonvariousdiseases,such asbreast, colon, duodenal, rectal,andpancreascancer, hypertension, diabetes, Alzheimer's etc. byextensive researches.Teaiscomposedofvariouscompounds. Most ofitsantioxidant effects areoftenattributedto epigallocatechingallet (EGCG). Therefore, necessity ofEGCG extraction and producing extractofantioxidant is obvious. In this study, the extraction of EGCG from green tea was investigated by modified supercritical CO 2 and Soxhletextraction with constant volume of co-solvent (1 ml), 25 min of static time and 0.674 mm of average particle size.Design of experiment carried out with response surface methodology (RSM) using Mini Tab software. The operating temperature (40-60 by step 5°C), the operating pressure (10-30 by step 5Mpa), the dynamic extraction time (40-120 by step 20min), and the flow rate of CO 2 (0.5-1.7 by step 0.3ml/min) have been considered as operating variables.Response surface analysis verified that the data were adequately fitted to second-orderpolynomial model.The linear and quadratics terms of temperature, pressure, CO 2 flow rate, and dynamic time, as well as the interaction between pressure-temperatureand dynamic time-temperature had significant effects on the proposed model ofEGCG recovery based on coded variables. R 2 and modified R 2 of the model are 98.40% and 96.99%, respectively. It was predicted that the optimum extraction conditions within the experimental ranges would be the extraction pressure of 19.29MPa, temperature of 43.7 , flow rate of 1.5 ml/min, and extraction time of 106 min with recovery of 0.462. Moreover, in the present study, a mathematical modeling for EGCG extraction from green teawas performedby modified supercritical carbon dioxide based on the differential mass balance. Density and viscosity of dense gases mixture was obtained by Peng-Robion (PR) equation of state with the van der Waals mixing rules and Chung et al., respectively.Mathematical model parameters are including effective pore diffusivity, film mass transfer coefficient, axial dispersion, anddistribution coefficient.The first three parameters were obtained fromempirical equations and the distribution coefficient between solid and solvent has been determinedby thermodynamic modeling of solubilities.Indicated by obtained results, the mathematical model is able to predict the experimental data with acceptable accuracy and R 2 is 98%.The main process conditions which must be determined to maximize the extraction recovery are temperature, pressure, flow rate of CO 2 , and dynamic extraction time. These were optimized by genetic algorithm. The optimal operating conditions were observed at 41.2 ?C, 19.79 MPa, 1.7 ml/min, and 116.3 min (dynamic time) to achieve 0.447 recovery.There was good agreement between two methods of optimization (genetic and RSM). Finally, neural network modeling was done by three hidden layers with16, 10 and 8neuron, respectively.The resultsshowedthe truly trained network and very good compatibility betweenthe neural networksand experimental data for EGCG recovery. Key Words Epigallocatechingallet, Supercritical extraction, Co-solvent, Response surface methodology,Mathematical modeling,Genetic algorithm, Neural network.