This paper proposes an intelligent control technique for fuel injection control of Compressed Natural Gas (CNG) engines. Recurrent Neuro-Fuzzy Networks are used to estimate and control air to fuel ratio (AFR) of CNG engines. To reasonably handle such a complicated control problem, a precise experimental test has been done on a real CNG fuelled vehicle and the process input output data have been collected by running the vehicle in transient conditions. To determine the proper amount of gas to be injected, a controller has been designed based on nonlinear inverse dynamics of AFR. The results show that the predicted results are in line with the measured fuel injection commands produced by the real electronic control unit (ECU). This evaluated and validated the efficiency of the controller. The control strategy has the advantage that control actions can be calculated analytically, avoiding the costly and time-consuming calibration efforts required in conventional fuel injection control strategies.