Although membrane-based separations became tailor-made for several applications in recent years, they still suffer from limited separation performance, i.e. the relation between permeability and selectivity and are also highly affected by fouling processes. Fouling can be generally divided into three categories, namely biofouling, colloidal fouling (especially by natural organic matter) and inorganic fouling (scaling). Besides operational conditions as well as feed water and foulant characteristics, membrane properties are a crucial factor for fouling mechanisms. Despite that many membrane functionalizations show improved anti-fouling behavior, their application is still limited to lab-scale or implementation during membrane fabrication. This project therefore focuses on surface-selective, reactive anti-fouling hydrogel coatings, which were applied in situ to nanofiltration membranes. To obtain such protective layer systems, first free radical polymerization of a zwitterionic copolymeric building block, which can be crosslinked to a hydrogel on surfaces by using a second crosslinkable functionality, was established with molecular weights in the range of 60-100 kDa. The gelation reaction of zwitterionic copolymer was initiated and studied in presence of radical initiator system (APS and TEMED) in free bulk. Critical gelation concentration was found to be above 5 wt.% and gelation time was in the range of 1-25 minutes, decreasing with higher polymer and initiator concentration. To apply such a zwitterionic hydrogel coating onto membranes, concentration polarization was utilized. According to the film model, filtration through a membrane will lead to a concentration gradient of retained substances, resulting from a balance of convective and diffusive mass transport. Thus, dead end filtration of crosslinkable building blocks and redox-initiator system, which were rejected by nanofiltration membranes (NF90 and NF270), lead to an increase of concentration at the membrane surface and enabled modification of membranes when critical concentrations were exceeded at the membrane surface. Additionally, the extent of concentration polarization during modification was modelled in order to discuss gelation in boundary layer and allow comparison to free bulk gelation conditions. Furthermore, the gelation reaction at membrane surface and its impact on membrane performance was studied in dependence of polymer and initiator feed concentration as well as filtration/reaction time. First, below a critical APS feed content (APS<0.06 wt%) gelation wasiv not initiated. Moreover, results indicate that at low polymer feed concentrations (>0.005 wt.%) only dense gelation in NF90 membrane hot-spots may occur, causing strong decline of permeability (~50%) without showing beneficial anti-fouling properties. Above this threshold concentration, hydrogel layers covered membrane surface completely. Furthermore, hydrogel layer thicknesses could be adjusted in NF270 membrane modification by higher polymer concentration and filtration time. For example, modifications performed below 15 minutes, resulted in layer thicknesses thinner than 100 nm and decreased membrane permeability only up to about 15%. Increasing filtration time to 40 minutes, lead to hydrogel layer thicknesses of approximately 2µm and a reduction in permeability up to 80%. Finally, hydrogel modified NF270 membranes showed higher permeability during protein filtration experiments compared to unmodified membranes.