Abstract In recent years, the research community devoted many resources to define accurate methodologies to model the real physics behind turbulent combustion. Such effort aims at reducing the need for case-by-case calibration in internal combustion engine simulations. In the present work two of the most widespread combustion models in the engine modelling community are compared, namely ECFM-3Z and G-equation. The interaction of turbulent flows with combustion chemistry is investigated and understood. In particular, the heat release rate characterizing combustion, and therefore the identification of a flame front, is analysed based on flame surface density concept rather than algebraic correlations for turbulent burn rate. In the first part, spark-ignition (S.I.) combustion is simulated in an optically accessible GDI single-cylinder research engine in firing conditions. The turbulent combustion regime is mapped on the Borghi-Peters diagram for all the conditions experienced by the engine flame, and the consistency of the two combustion models is critically analysed. In the second part, a simple test case is defined to test the two combustion models in an ideally turbulence-controlled environment: this allows to fully understand the main differences between the two combustion models under well-monitored conditions. and results are compared against experimental databases of turbulent burn rate for wide ranges of Damkohler (Da) and Karlovitz (Ka) numbers. The joint experimental and numerical study presented in this paper evaluates different approaches within the unified flamelet/non-flamelet framework for modelling turbulent combustion in SI engines. It also indicates guidelines for reduced calibration effort in widespread combustion models.