Batch sorbitol to sorbose bioconversion by Acetobacter suboxydans using initial sorbitol concentration (S-0 = 100 g/L) yielded a productivity of 10.11 g/L.h and 98.6% conversion in 10 h time. The batch kinetics was then used to develop an unstructured mathematical model. Model parameters were identified using a nonlinear regression technique assisted by a computer program which minimized the deviation between the model predictions and actual batch experimental data. F test indicated 99% confidence on the prediction of model using optimized parameters. The batch model was eventually extrapolated to identify nutrient feeding strategies to maintain constant noninhibitory sorbitol supply and eliminate substrate limitation for fed-batch fermentation in order to improve the sorbose productivity. The adequacy of the fed-batch model was established by excellent agreement between experimental data and model simulation (except towards the end of fermentation).