The two-stage anaerobic wastewater pre-treatment was modelled and controlled. The biological state of the reactors could be predicted using a fuzzy logic system and based upon this, proper control actions were taken automatically in order to avoid an overload. The system was designed to handle very strong fluctuations in the concentration of the substrate and the volumetric loading rate. Hydrogen concentration together with methane concentration, gas production rate. pH and the filling level of the acidification buffer tank were used as input variables for the fuzzy logic system. The manipulated variables were the flow rate from the acidification buffer tank into the methane reactor, the temperature and pH of both reactors, the circulation rate of the fixed bed reactor, back flow from the methane reactor into the acidification, and the control of the feed into the acidification buffer tank. The developed control system was successfully tested on a fully automated lab scale two-stage anaerobic digester. Different types of wastewater from food processing industries were successfully applied. Even a restart of feeding with very high COD concentrations (100 gl(-1) after several days of stand by was handled successfully. Effluent concentrations could be kept low without using TOC, COD or equivalent measurements.