The extraction of marigold (Calendula officinalis) oleoresin with supercritical carbon dioxide SCCO2 was carried out in a small scale extraction vessel (2.7×10-4 m 3) studying different extraction pressures (14, 24 and 34 MPa) and different flow rates (15, 30 and 45 g•min-1) at 313 K. Then, using semi-empirical engineering scaling criteria (constant solvent lineal velocity or constant solvent residence time) and the Broken and Intact Cell (BIC) model, the scaling up to larger extraction vessels (1.35×10-3 m 3 and 5.16×10-3 m 3) was theoretically investigated. According to the BIC model, by keeping constant the CO2 residence time in the different size vessels a good reproduction of the kinetic behavior should be obtained. Nevertheless, experimental results did not confirm model predictions, and in fact none of the scaling criteria studied resulted adequate in the marigold supercritical extraction scaling up. Thus, using all the experimental overall extraction curves obtained, a new specific correlation was developed between the Schmidt number (Sc), CO2 mass flow, bed geometry and the supercritical mass transfer coefficients kYA with a good fit (R 2 = 0.9767) for scaling up the supercritical extraction of marigold.