Air-ratio relates closely to pollution reduction and fuel efficiency improvement among all of the engine control variables. Maintaining the air-ratio to be the stoichiometric value can ensure the maximum efficiency of the three-way catalytic converter so that minimizing the engine emission. The thesis presents a new model predictive control (MPC) algorithm for air-ratio regulation based on relevance vector machine (RMV). The control algorithm has been implemented on a real car to test. Experimental results show that the control performance of the relevance vector machine model predictive controller (RVMMPC) is superior to typical neural network MPC, decremental least-squares support vector machine MPC and conventional proportional-integral (PI) controller in production cars. Therefore, the RVMMPC is a potential scheme to replace PI controller in the automotive ECU for engine air-ratio control.