Supercritical CO2 (scCO(2)), used in the food industry as a water extraction agent, requires dehydration units for regeneration. The present study assesses the economics of membrane-based dehydration of scCO(2). In contrast to earlier studies, the contribution, next to the membrane, also the contributions of the mass transfer resistances of the feed and permeate boundary are included, which have a dominant effect on the final process design and economics. In addition, our work also extrapolates the process to industrial scale evaluating different configurations and process conditions. Specifically, the contribution of the membrane and membrane unit costs is discussed in more detail. Including the mass transfer resistances of feed and permeate boundary layer reduces the water flux across the membrane up to a factor 150, implying a larger required membrane surface area for a given water removal rate, and thus higher costs. Using a SPEEK-based membrane, the total drying costs, normalized for the amount of water removed, minimize around a skin layer thickness of 1 mu m, i.e., not too thin to permeate and thus spill too much CO2 and not too thick to hamper the H2O flux. Because the feed boundary layer dominates water transport, conditions that minimize its thickness reduce total costs. A reduction of the feed boundary layer dominance can be achieved by adjusting channel height, cross-flow velocity and the density and viscosity of scCO(2), the latter two by increasing the operational temperature from 45 to 65 degrees C (at 130 bar). Compared to the benchmark zeolite process currently available, the membrane-based process for drying scCO(2) outlined and optimized in the present study results in a 50% saving of total drying costs. These savings can be achieved by using a dense polymeric membrane with a H2O permeability of at least 10,000 Barrer and a CO2 permeability of at most 10 Barrer.