RadarDistill is a novel knowledge distillation (KD) method that enhances radar-based 3D object detection by leveraging LiDAR data. The method introduces three key components: Cross-Modality Alignment (CMA), Activation-based Feature Distillation (AFD), and Proposal-based Feature Distillation (PFD). CMA improves radar feature density by applying dilation operations, AFD selectively transfers knowledge based on active regions in LiDAR features, and PFD guides radar networks to mimic LiDAR features within object proposals. RadarDistill achieves state-of-the-art performance on the nuScenes dataset, with a 20.5% mAP and 43.7% NDS for radar-only detection, and significantly improves camera-radar fusion models. The method demonstrates that transferring knowledge from LiDAR to radar can enhance radar features, leading to better object detection and localization. The study highlights the importance of CMA in resolving inefficient knowledge transfer between radar and LiDAR point clouds. RadarDistill also introduces two novel KD methods, AFD and PFD, which bridge the gap between radar and LiDAR features at different levels. The proposed framework achieves significant performance improvements in both radar-only and camera-radar fusion scenarios.RadarDistill is a novel knowledge distillation (KD) method that enhances radar-based 3D object detection by leveraging LiDAR data. The method introduces three key components: Cross-Modality Alignment (CMA), Activation-based Feature Distillation (AFD), and Proposal-based Feature Distillation (PFD). CMA improves radar feature density by applying dilation operations, AFD selectively transfers knowledge based on active regions in LiDAR features, and PFD guides radar networks to mimic LiDAR features within object proposals. RadarDistill achieves state-of-the-art performance on the nuScenes dataset, with a 20.5% mAP and 43.7% NDS for radar-only detection, and significantly improves camera-radar fusion models. The method demonstrates that transferring knowledge from LiDAR to radar can enhance radar features, leading to better object detection and localization. The study highlights the importance of CMA in resolving inefficient knowledge transfer between radar and LiDAR point clouds. RadarDistill also introduces two novel KD methods, AFD and PFD, which bridge the gap between radar and LiDAR features at different levels. The proposed framework achieves significant performance improvements in both radar-only and camera-radar fusion scenarios.