Fuel film dynamics in the intake manifold are considered to develop air fuel ratio (AFR) control strategy with on-line system identification for a V2 engine in this paper. A1000 cc four-stroke two-cylinder, water-cooled port injection SI engine is used as the target engine to develop the engine model in Matlab/Simulink. The model which consists of charging, fueling, combustion, friction, and engine rotational dynamics is used to verify the proposed AFR control. Since the fuel film dynamics changes with different engine operating conditions, the fuel film parameters are often listed as look-up tables for fuel film dynamics calculation in the conventional AFR control. However, those parameters might be inaccurate during transient engine operation. Different intake port temperature will affect the accuracy of those fuel film parameters as well. In order to solve this problem, recursive least square (RLS) is used to identify those parameters on-line. Kalman filter is utilized to estimate the AFR using a narrow-band oxygen sensor. Model predictive control is used to design the proposed AFR controller using the identified parameters of fuel film dynamics and estimated AFR. Simulation results show that those parameters identified using RLS are more accurate to describe the transient fuel film dynamics than that from look-up tables. In addition, using the proposed AFR control strategy, AFR can be controlled at the desired value both in steady state and transient engine operation.