This paper presents a design for a fuzzy gain tuning mechanism dealing with the problem of unbalanced vibration problem in an active magnetic bearing (AMB) system. For the purpose of enhancing the performance of the AMB system, we replace the conventional proportional-integral-derivative (PID) controller with a self-tuning fuzzy PID-type controller (FPIDC). The shaft displacement and the unbalanced forces of the rotor are evaluated by model-based observation. If there are model uncertainties in the rotor system or nonlinearities in the magnetic bearing system, this observer may not work well at any operating speed. A fuzzy gain tuner is added to adjust the actuating signal of the self-tuning FPIDC in order to overcome the disturbances and suppress the unbalancing vibration. The experimental results show that the proposed scheme allows for a remarkable improvement in reducing vibration in an unbalanced AMB system as well as demonstrate an efficient reduction in the shaft displacement of the rotor. (C) 2008 Elsevier Ltd. All rights reserved.