In order to solve the environmental pollution problem and the energy depletion problem in recent years, a control technology that improves the quality of engine in automobile is demanded. However, thanks to the developments of electrical and electronic mounting technology, advanced control of the power train has become possible. This paper presents an application of "Large scale database-based Online Modeling (LOM)" for the cold starting control of SI engine. LOM is a local modeling technique based on the database to predict and control a large-scale process. The intake air mass flow in cylinder is predicted in order to reach the desired engine speed in cold starting by using LOM, and adequate fuel injection quantity is derived from the intake air mass flow and the fuel injection model.