Abstract: A nonlinear two-stroke engine model was constructed using a mathematical tool of stepwise regression.Taylor expansion method was used around an equilibrium point during the linearizing process,and the linearized model was reached.A method for designing robust proportional-integral(PI) controller was presented.The problem of parameter designing of the robust PI controller was solved with linear matrix inequalities(LMIs) approach.The controller switching was designed based on the idea of parallel distributed compensation(PDC) to Takagi-Sugeno fuzzy model.Simulation and experiments on the engine-dynamometer show that the proposed controller can keep the engine speed in 2.3% under 11% disturbance and can achieve the anticipated robust performance.