Trailing edge slot film cooling is a widely used active cooling scheme for turbine blade trailing edges. Current Reynolds-Averaged Navier–Stokes (RANS) models are known to significantly overpredict the adiabatic effectiveness of these configurations. It is shown that this overprediction is due in part to the breakdown of the Reynolds analogy between turbulent shear stress and scalar transport in the near wall region. By examining previously reported direct numerical simulation (DNS) results for a wall-mounted cube in cross flow, it is seen that in a flow with a significantly perturbed outer boundary layer, the turbulent diffusivity is not as strongly damped as the turbulent viscosity in the viscous sublayer and buffer layer of the boundary layer. By removing the Van Driest damping function from the length scale model for the turbulent diffusivity, more accurate turbulent diffusivity predictions are possible. This near wall correction is applied to trailing edge slot film cooling flows and it is demonstrated that the predictive accuracy of the RANS models is significantly enhanced. Detailed comparisons between RANS results and experimental datasets for 15 different cases demonstrate that this correction gives significant improvement to the accuracy of the RANS predictions across a broad range of trailing edge slot film cooling configurations.