The study objective is to assess the impact of cloud-permitting resolution and improved representation of multiscale processes on the ability to predict rapid intensification (RI) and structure of Phailin (2013), one of the strongest tropical cyclones (TCs) over the Bay of Bengal. The state-of-the-art Hurricane Weather Research and Forecasting (HWRF) modelling system is used with two different configurations. The first configuration uses a static domain of 27 km grid size with a movable nested domain of 9 km grid size (hereafter H2D). The second configuration has an additional movable nested domain of 3 km grid size (known as H3D) to resolve meso-and vortex-scale features respectively.The results clearly show the ability of the H3D system at cloud-permitting resolution (3 km) in predicting the TC movement, intensity and structure. The storm-to-vortex scale interaction in H3D allowed for better prediction of large-scale wind flow, low-level wind asymmetry and PV tendency, and provided insight to improve track predictions. The vortex depth is another important factor and the shallow vortex in the H2D run interacted differently with the large-scale environment and resulted in large track and intensity errors. Substantial gains are noticed in RI and structure prediction, mainly due to better simulation of diabatic heating, strong inflow, and moisture distribution in H3D, where the intensity errors are ≤11 knots (5.6 m s −1 ) up to the 72 h forecast, and up to 40 knots (20.5 m s −1 ) in the H2D version. The upper-level warming is well resolved in the H3D as compared to the H2D run. In summary, this study highlights the need for considering multiscale interactions and improved physics along with high-resolution initialization to resolve convective processes in the vortex and to realistically predict track, structure, and intensity changes. is defined as an increase of 30 knots (15.4 m s −1 ) in 3 min sustained maximum wind speed in a 24 h period.