Background The global market for sweeteners is increasing, and the food industry is constantly looking for new low-caloric sweeteners. The natural sweetener 5-keto-d-fructose is one such candidate. 5-Keto-d-fructose has a similar sweet taste quality as fructose. Developing a highly efficient 5-keto-d-fructose production process is key to being competitive with established sweeteners. Hence, the 5-keto-d-fructose production process was optimised regarding titre, yield, and productivity. Results For production of 5-keto-d-fructose with G. oxydans 621H ΔhsdR pBBR1-p264-fdhSCL-ST an extended-batch fermentation was conducted. During fructose feeding, a decreasing respiratory activity occurred, despite sufficient carbon supply. Oxygen and second substrate limitation could be excluded as reasons for the decreasing respiration. It was demonstrated that a short period of oxygen limitation has no significant influence on 5-keto-d-fructose production, showing the robustness of this process. Increasing the medium concentration increased initial biomass formation. Applying a fructose feeding solution with a concentration of approx. 1200 g/L, a titre of 545 g/L 5-keto-d-fructose was reached. The yield was with 0.98 g5-keto-d-fructose/gfructose close to the theoretical maximum. A 1200 g/L fructose solution has a viscosity of 450 mPa∙s at a temperature of 55 °C. Hence, the solution itself and the whole peripheral feeding system need to be heated, to apply such a highly concentrated feeding solution. Thermal treatment of highly concentrated fructose solutions led to the formation of 5-hydroxymethylfurfural, which inhibited the 5-keto-d-fructose production. Therefore, fructose solutions were only heated to about 100 °C for approx. 10 min. An alternative feeding strategy was investigated using solid fructose cubes, reaching the highest productivities above 10 g5-keto-d-fructose/L/h during feeding. Moreover, the scale-up of the 5-keto-d-fructose production to a 150 L pressurised fermenter was successfully demonstrated using liquid fructose solutions (745 g/L). Conclusion We optimised the 5-keto-d-fructose production process and successfully increased titre, yield and productivity. By using solid fructose, we presented a second feeding strategy, which can be of great interest for further scale-up experiments. A first scale-up of this process was performed, showing the possibility for an industrial production of 5-keto-d-fructose.