Laminarization is an important topic in heat transfer and turbulence modeling. Recent studies have demonstrated that several well-known turbulence models failed to provide accurate prediction when applied to mixed convection flows with significant re-laminarization effects. One of those models, a well-validated cubic nonlinear eddy-viscosity model, was observed to miss this feature entirely. This paper studies the reasons behind this failure by providing a detailed comparison with the baseline Launder–Sharma model. The difference is attributed to the method of near-wall damping. A range of tests have been conducted and two noteworthy findings are reported for the case of flow re-laminarization.