The paper describes dynamic modelling of an industrial batch fermentation process for production of an antibiotic. The aim of the study is to find out whether the established modelling techniques can be applied to an industrial process with a complex substrate, where only approximate knowledge about the process is available and without costly additional measurements and experimentation. A nonlinear state-space model was built, and the model can successfully reproduce the process variable profiles of a typical batch run in an open-loop simulation. The process model is used for state estimation of key process variables, like biomass and product concentrations, which are not in the set of measured variables due to high measurement costs.