Methods based solely on images often overly rely on inter-frame information and lose crucial motion cues, resulting in performance bottlenecks for motion deblurring. Event cameras with high temporal resolution offer a promising breakthrough in the ability to effectively capture motion information. However, most existing event-based methods operate only in the spatial domain, neglecting cross-dimensional interactions and the frequency differences between event data and blurred images, leading to suboptimal fusion and limited effectiveness in handling complex blur. To address these challenges, we propose Event-driven motion deblurring based on Multidimensional Interaction and Frequency-domain Separation (EMIFS). This method mitigates information loss during cross-modal fusion and improves deblurring performance. Specifically, we design a multidimensional interaction attention module to enhance the model’s perception of complex blurred regions and improve detail restoration. In the fusion stage, we introduce a frequency-domain separation based multimodal bidirectional fusion module, enabling effective integration of complementary information from events and blurred images through frequency separation and recombination. Extensive experiments on the public GoPro and REBlur datasets demonstrate that EMIFS achieves superior deblurring performance compared to state-of-the-art methods, with PSNR gains of 0.4 dB and 0.19 dB, respectively.