This study focuses on 3D conjugate heat transfer CFD simulations of building envelope samples with air cavities and thermal bridges. Such simulations aim at supporting certification of wall assemblies in terms of effective thermal resistance. Conduction, convection and radiation heat transfer modes are simulated on a conformal mesh of the assembly components and air cavities, allowing to solve explicitly the 3D heat flow paths. Numerical modeling was primarily based on procedures defined in ASHRAE-1365-RP (Thermal Performance of Building Envelope Details for Mid- and High-Rise Buildings), including contact resistances and interior and exterior boundary conditions. However instead of using generic heat transfer coefficients for the cavities as defined in 1365-RP, the flow is explicitly resolved using a CFD solver, allowing to remove empiricism when treating different cavity shapes and sizes. A base case model (8’x8’ sample of an insulated steel stud wall assembly) has been compared against a corresponding guarded hot-box thermal test (ASTM C1363) for validation purposes. The comparison between simulation and physical testing shows good correlation, within 1.5% for the R-value prediction. The comparison highlights the importance of considering the stud tracks in the model even when looking at clear field R-values because of their intrinsic contribution in the formation of in-plane convection cells between studs. It is also observed that modeling the screws of the wall assembly – which is not typically done - improves the prediction when comparing with experimental data. Representing explicitly convection and radiation heat transfer in cavities allows to characterize their thermal resistance regardless of their shape or size. This provides a better understanding of the 3D heat flow paths and allows to target the most impactful changes for thermal performance of building envelopes.