Recently, the possibility of producing fructosylated chondroitin from the capsular polysaccharide of Escherichia coli O5:K4:H4, in fed-batch and microfiltration experiments was assessed on a 2 L bioreactor. In this work, a first scale-up step was set on a 22 L membrane reactor with modified baffles to insert ad hoc designed microfiltration modules permanently inside the bioreactor vessel. Moreover, the downstream polysaccharide purification process, recently established on the AKTA cross-flow instrument, was translated to a UNIFLUX-10, a tangential flow filtration system suitable for prepilot scale. In particular, the microfiltered permeates obtained throughout the fermentation, and the supernatant recovered from the centrifuged broth at the end of the process, were treated as two separate samples in the following ultrafiltration procedure, and the differences in the two streams and how these affected the ultrafiltration/diafiltration process performance were analysed. The total amount of K4 capsular polysaccharide was about 85% in the broth and 15% in the microfiltered permeates. However, the downstream treatment was more efficient when applied to the latter. The major contaminant, the lipopolysaccharide, could easily be separated by a mild hydrolysis that also results in the elimination of the unwanted fructosyl residue, which is linked to the C-3 of glucuronic acid residues. The tangential ultrafiltration/diafiltration protocols developed in a previous work were effectively scaled-up, and therefore in this research proof of principle was established for the biotechnological production of chondroitin from the wild-type strain E. coli O5:K4:H4. The complete downstream procedure yielded about 80% chondroitin with 90% purity. (c) 2012 American Institute of Chemical Engineers Biotechnol. Prog., 28: 10121018, 2012