Low-light images often have poor visual quality, with noise amplified when brightness is enhanced. Effective enhancement requires balancing brightness, noise suppression, and structural preservation. Despite the importance of structural information in low-light image enhancement (LLIE), many existing methods neglect it, leading to images lacking detail. This paper introduces the I3En network, a simple yet effective approach that integrates structural clarity through a multi-level iterative restoration strategy and a convolution and self-attention-based multi-scale feature extraction module. The I3En network emphasizes structural information, enhancing visual quality and outperforming state-of-the-art methods in quantitative metrics like SSIM by 3.02%, 4.30%, and 7.45% on the LOLv1, LOLv2-real, and LSRW datasets, respectively. The results highlight I3En’s transformative impact on LLIE, prioritizing structural fidelity and visual clarity for superior enhancement outcomes. Code is available at: https://github.com/hanxing-go/I3En .