This paper presents a new switching control scheme for an active magnetic bearing (AMB) system using self-tuning fuzzy proportional-integral-derivative (PID) control. The research process consists of three stages. First, four types of self-tuning fuzzy PID-type controllers (FPIDCs) consisting of two most commonly used fuzzy inference systems: Mamdani and Takagi-Sugeno types, and two efficient parameter adaptive methods: function tuner and relative rate observer, are used to control a highly nonlinear AMB system, respectively. Hence, there are two kinds of FPIDCs can be obtained by comparing experimental results of these tests: one has the fastest transient response and the other has the minimum steady-state error. Next, the switching-type self-tuning FPIDC is proposed by combining the two kinds of FPIDCs. Namely, the AMB system is dominated by the scheme with the fastest transient response when the rotor is at rest and by the one with the best steady-state performance when the rotor is in rotation. Finally, experimental results demonstrate that the proposed switching-type self-tuning FPIDC performs better overall performance than the other self-tuning FPIDCs, particularly when controlling an AMB system.