The fuel injection timing is one of the most important operating parameters that affect the atomization, mixture formation and combustion which determines the performance and emissions of a gasoline engine. Optimizing the injection timing will improve the performance of the engine to a large extend. Towards this end artificial neural-network (ANN) technique using Levenberg-Marquardt (LM) training algorithm is used to train and optimize the fuel injection timing of a single cylinder, four-stroke gasoline engine. Experimental studies have been carried out to obtain training as well as test data. For various engine speeds between 700 and 5000 rpm and for different manifold absolute pressures, fuel injection timing was measured by conducting experiments. The experimental data set generated is used to train the neural network to arrive at the optimized performance of the engine. The optimized fuel injection timing arrived at from ANN is validated by conducting experiments again on the same single cylinder gasoline injected engine from where the initial set of data were obtained. The ANN predicted results are found to be within good acceptable limits and the results show close agreement between predicted and experimental values. From this study it is concluded that for optimizing engine performance with respect to injection timing ANN with LM algorithm can be advantageously used because it saves time and cost.