Energy costs involved in the recovery of bio-value-added products from very dilute solutions are much higher than the costs of raw materials and fermentation and are crucial to make the production economically viable. The aim of the study is to show that micro-filtration can help recovering the enzyme used in the fermentation of tobacco samples to create bioethanol, once it is an energy efficient process. The “experimental” (EX) samples were the whole plant. The “by-product” (BY) consisted mainly on the stem. The samples were cut and frozen immediately after harvesting. The hydrolysate, made from EX and BY samples, was prepared in a 2L fermentation unit (30°C±0.2, pH 4.5±0.1) (Labfors Minifors, Belgium). The enzyme was endo-1,4-β xylanase (Sigma Aldrich) from Trichoderma longibrachiatum. Separation was carried out by a stirred cell device (400cm3) with a 0.004534m2 polyether-sulfone membrane with an MWCO (0.1µm). The efficiency of the process was expressed by the retention, and protein quantity determined by the Kjeldhal method. The initial values of the filtration were greater for the BY samples (31, 76 L/m2h) than for the EX samples (24,06 L/m2h). Three different membrane resistance values were calculated (membrane resistance, gel layer on the surface of the membrane resistance and fouling resistance). The values for the BY samples were 3,75x1013m-1, 1,48x1012m-1, 9,34x1011m-1, respectively, and for the EX samples 3,09x1013m-1, 1,48x1012m-1, 9,34x1011m-1. The value of the protein retention of the EX samples (17,13%) was higher than the BY samples (4,49%). It can be concluded that the enzymatic hydrolysate of the cellulose content of EX contain several smaller fragments than the BY samples. The EX samples contain the whole plant, so they have less thick cellulose fibers. This study shows that the membrane filtration could help recovering the enzyme, which was used in the fermentation of tobacco samples to create bioethanol.