Nowadays, the development of a new engine is becoming more and more complex due to conflicting factors regarding technical, environmental and economic issues. Hence, the experimental activity has to comply with the above complexities, resulting in increasing cost and duration of engine development. For this reason, the simulation is becoming increasingly important, thanks to its lower financial burden, together with the need of an improved predictive capability. Among the other numerical approaches, the 1D models represent a proper compromise between reliability and computational effort, especially if the engine behavior has to be investigated over a number of operating conditions. The combustion model has a key role in this contest and the research of consistent approaches is still on going. In this paper, two well-assessed combustion models for Spark Ignition (SI) engines are described and compared: the eddy burn-up theory and the fractal approach. Both are embedded in the commercial software GT-Power under the form of “user routine”. The main aim of the paper is a detailed appraisal of the above combustion models, carried out with reference to three SI engines, both at full and part load operations. In a first stage, a sensitivity analysis is performed to investigate the effects of each tuning constant. Only for the fractal model, thanks to selective influence of the tuning constants on specific phases of the combustion process, a simplified tuning procedure is conceived. The eddy burn-up approach does not allow a standardized tuning method because of a widespread effect of the constants over the whole combustion process. After a tuning, where a fixed set of constants is identified for each engine, the predictive capabilities of the combustion models are compared under various operating conditions, in terms of characteristic combustion events and burn rate profiles. Both present a good consistency for all the tested engines, at full and part loads. The fractal approach denotes a slightly higher accuracy with respect to the eddy burn-up approach, especially in terms of burn rate shape and capability to sense different turbulence levels. Finally, the generality of the combustion models is investigated assigning a unique tuning for all the engines. Both approaches show a satisfactory capability to furnish consistent results also without a dedicated tuning. Summarizing, the fractal and eddy burn-up combustion models proved a similar satisfactory reliability and generality for various engines and operating conditions. However, the fractal approach denoted an easier tunability and a slightly higher accuracy.