<p>Hurricanes can cause significant damage to society and infrastructure. However, the understanding of planetary boundary layers (PBLs) in hurricanes is limited due to insufficient available measurements and a small number of conducted high-resolution large-eddy simulations (LESs) of hurricane flows. While many issues were found in the default PBL schemes of mesoscale models for hurricane forecasts, the role of the sub-grid scale (SGS) parameterizations in coarse LESs for real hurricane forecasts has not been well established. This study aims to bridge this gap by simulating five major real hurricanes using the weather research and forecasting (WRF) model. In total, 33 WRF-LESs were performed by varying the eddy viscosity, grid resolution, and hurricane cases. Reducing the eddy viscosity in the 1.5-order turbulent kinetic energy (TKE) SGS stress model of WRF-LES intensified the hurricanes, decreased the elevation of the maximum low-level jet, reduced the size of eddies, and enhanced horizontal mixing and fluctuations. It also appeared to outperform the surface wind intensity forecasts by ~ 9% and minimum sea-level pressure predictions by ~ 29% on average for five hurricanes compared to the default eddy viscosity coefficient in the 1.5-order TKE SGS model. The horizontal turbulent fluxes of the decreased SGS cases also showed a better agreement with high-resolution LESs. These results suggest that the 1.5-order TKE SGS model may not adequately account for turbulence suppression effects caused by strong rotation in hurricanes. Such effects could be significant for low-resolution LESs that rely more on the SGS parameterizations. This study’s findings pertain to the considered WRF-LES model. These results provide new insights into the impacts of 1.5-order TKE SGS models on low-resolution LESs for real hurricane forecasts.</p>