The ethanol concentration in batch cultivation with the yeast S. cerevisiae was predicted on-line using a gas sensor array. Head space samples were pumped past the gas sensors array every five minutes for 10 seconds and the voltage changes of the sensors were measured. For the calibration procedure no off-line sampling was used. Instead, a theoretical model of the process has been applied to simulate the ethanol production at any given time. However, the kinetic parameters of the simulation model are unknown at the beginning of the calibration. It will be demonstrated that these kinetic parameters of the theoretical process model can be acquired from the response of the gas sensor array alone. The calculated parameters result in a simulation model that is at least as accurate as a model whose parameters are acquired by least squares fitting to off-line measurements. The root mean square error of calibration as well as the percentage error for validation sets was below 0.2 g/L and 7 %, respectively. The obtained results indicate that, the model-based calibrated gas sensor array can be a cheap alternative to other tools that are used for monitoring yeast cultivations such as spectroscopy based methods.