Reynolds-averaged simulations (RAS) and large-eddy simulations (LES) are performed of a model strut fuel injector for scramjet applications at hypersonic flow conditions. The strut injector and flow conditions have been also studied experimentally. In the experiments, planar laser-induced fluorescence (PLIF) data and the instream pitot pressure, total temperature, and gas sampling measurements were obtained at several locations downstream of the strut. Mixing efficiency is then computed from simulation and experimental data. Because the simulation domain is relatively large, the domain upstream of the fuel injection plane is truncated to reduce the simulation domain size. The inflow boundary conditions for the downstream mixing region are obtained from the outflow plane of the upstream RAS and additional insight from PLIF. Although this approach is common in practice and reduces the computational cost of LES, specifying the inflow conditions accurately for LES becomes challenging. For RAS, typical two-equation linear eddy viscosity and diffusivity models are used. Such simulated mixing flowfield exhibits a strong dependence on the turbulent Schmidt number and the turbulence model, which are adjusted until the mixing efficiency mimics the experimental data. When the experimental data are absent, LES have been proposed as a surrogate for the experiments that could provide the data needed to “calibrate” the RAS. Using LES only as a surrogate is motivated by the significant computational cost (CPU, data storage, and time) of LES as compared to RAS, making LES prohibitive for use in most engineering applications and specifically for parameter exploration or optimization. In the current work, despite uncertainty in the inflow boundary conditions, LES still demonstrates greater fidelity to capture mixing flow physics as compared to “calibrated” RAS. This demonstration is achieved by comparing the mixing efficiency obtained from the experiments, LES, and RAS, where the latter has been evaluated for a range of turbulent Schmidt numbers and turbulence models. In addition, the least squares fitting procedure has been applied to the LES data to obtain an estimate for the turbulent Schmidt number analytically. The work demonstrates a practical approach by which both the turbulent Schmidt number and the RAS turbulence model that together most faithfully reproduce the flow could be determined.