The cooling system for internal combustion engines removes waste heat from engine block and head and reject it to the environment to maintain the desired coolant temperature for enhanced performance. It’s known that cooling system consumes a portion of engine’s power, and for a better fuel economy and less pollutants emissions, it is important to develop and optimize the system for a minimum power and energy consumption. Over the latest years, virtual thermo-fluid analysis has been utilized as a simpler and less costly way to optimize and pre-select the best characteristics of a particular system. Considerable power to cooling can be reduced by correct design of system, in addition to cost reduction. Under this point of view, for a naturally aspirated and port fuel injection engine, the paper approaches a cooling system optimization using virtual correlated models, aiming in the reduction of the length of radiators and in the reduction of necessary fan power of the cooling system, which has a direct impact on global engine efficiency. Weight reduction was reached by optimization of radiator size and consequently, coolant volume, which impacts in a faster engine warm up due to less thermal capacity, and it has good impact in fuel economy due to less friction losses. Based on a design of experiment methodology, the studied parameters were varied in a specific range to attend the standards that guide the development of cooling systems, allowing to test many combinations of different parameters in a very fast time, when compared with experimental tests. The results data are displayed in an innovative way allowing to choose the best combination of characteristics that attends a particular vehicle and also providing an initial step to further studies of the impacts of the parameters in others vehicle’s sub-systems.