This paper addresses the problem of optimization of hydrogen production in continuous anaerobic digesters using a model predictive control (MPC) strategy. The process is described by a dynamic nonlinear model. The influent concentration of molasses together with the effluent substrate and product concentrations of acetate, propionate, butyrate and biomass were estimated by an asymptotic online observer from measurements of gas composition in H-2 and CO2 and gas flow rate. The observer was tested experimentally before to apply MPC online. The combined strategy (MPC and observer) was used in order to optimize a bioreactor of 2 L. The hydrogen production was increased by 75% up to 8.27 mLH(2) L-1 min(-1), using the influent flow rate as the main control variable while keeping the conversion of the influent concentration higher than 95% and maintaining the temperature at 37 degrees C and pH at 5.5.