Turbulent combustion modelling under high-pressure conditions is a key issue for the design of future aero or rocket engines. In the context of large-eddy simulations, the exact filtered equation of state (EoS) is generally approximated by an EoS directly computed from the Favre filtered quantities for species mass fractions and temperature. The soundness of this approximation is presently addressed through the analysis of laminar CH4-air and CH4-O-2 high-pressure premixed and non-premixed flames. Their computations were performed either with the ideal gas equation of state (EoS) or with a high-pressure package that gather a cubic EoS along with additional pressure-dependent transport parameters and thermodynamic relations. Various kinetic schemes are used for the computation of laminar premixed flames and a comparison with available experimental data is provided for flame speed (S-L). Laminar premixed CH4-O-2 flames exhibit micro metric flame thickness for pressures above 2.0 MPa as well as a high flame temperature (\textgreater3000 K). These flamelets were then filtered either with a Gaussian filter or through a beta-pdf technique. The resulting filtered profiles were used to approximate the filtered pressure, (p) over barT. A negligible error is observed for CH4-air flames when comparing (p) over bar with the exact filtered pressure. However, significant errors were found for high-pressure CH4-O-2 flames, that increase with the filter size. This behaviour is exacerbated for transcritical injections meaning that classical techniques that use tabulated thermochemistry methods in compressible codes must be revisited in that context. A correction to the tabulated chemistry approach coupled to a presumed subgrid pdf is finally proposed to comply with high-pressure considerations. (C) 2016 Elsevier B.V. All rights reserved.