Graphene has many outstanding properties which make it a prime candidate for new technology. At the current time it is very difficult and expensive to produce large sheets of graphene, but there are many applications where that is not necessary and smaller flakes of graphene can be used instead. A practical way of handling these graphene flakes is in a dispersion, especially a water-based dispersion have many benefits. Such a stable dispersion of functionalized graphene is produced, improved, and characterized in this project. An aqueous system that was developed in two previous M.Sc. theses, each determining a suitable graphene powder and stabilizer, was used as a starting point with the main purpose being to improve the yield. The method used to produce these dispersions can be described as sonicating graphene powder in a solution of water and stabilizer followed by centrifuging to remove un-dispersed graphene particles. Experiments were carried out examining the possibility of dispersing those previously undispersed graphene flakes, combining the stabilizer with several surfactants, optimizing the centrifuge speed and time, refining the sonication procedure with longer exposure time and cooling, narrowing the size-distribution of the original stabilizer through ultrafiltration, and removing excessive unbound stabilizer through ultrafiltration. Samples were characterized with UV-vis, SEM, TGA, Electrophoretic light scattering, and Laser diffraction spectroscopy. It was discovered that the yield from the graphene powder was heavily dependent on sonication time and centrifugation conditions. The gain from increasing sonication time showed that most, if not all, of the un-dispersed graphene flakes previously considered lost could in fact be dispersed. In an industrial setting any un-dispersed flakes could simply be added to the next batch. Reducing the centrifugation speed as well as time increased the concentration of graphene to more than twice as high, and that gain comes solely from the larger graphene flakes. Thusly the previous problem with a low yield was shown to have been caused by too little sonication and too much centrifugation. The particle size analysis did show a small reduction in flake size as the sonication time was increased, but when those dispersions were characterized in SEM they all formed even films with no discernable difference between them. Purifying the scaled up dispersions by removing excess stabilizer through ultrafiltration was performed to three different degrees, 0 %, 50 % and 95 %, for a total of three dispersions of 100ml. All three dispersions were shown to be highly stable, with no apparent reduction in graphene concentration over 5 weeks and a zeta potential averaging below -50mV. The TGA results reinforce the UV-vis results, proving that the purification worked as intended.