RetinexMamba is a novel architecture combining Retinex theory with Mamba for low-light image enhancement. It addresses the limitations of traditional Retinex methods and deep learning approaches like Retinexformer by integrating the computational efficiency of State Space Models (SSMs) and introducing an innovative illumination estimator and damage restorer. RetinexMamba replaces the Illumination-Guided Multi-Head Attention (IG-MSA) in Retinexformer with a Fused-Attention mechanism, improving model interpretability. The architecture features an Illumination Estimator (IE) and an Illumination Fusion Visual Mamba (IFVM), with the IFVM using an Illumination Fusion State Space Model (IFSSM) to achieve linear computational efficiency. The IFSSM employs 2D Selective Scanning (SS2D) and an Illumination-Fused Attention (IFA) mechanism to enhance feature extraction and attention calculation. Experimental evaluations on the LOL dataset show that RetinexMamba outperforms existing deep learning approaches in both quantitative and qualitative metrics, demonstrating its effectiveness in enhancing low-light images. The model achieves higher PSNR and lower RMSE compared to state-of-the-art methods, while maintaining better image quality and reducing color distortion and noise. Ablation studies confirm the model's superiority, with the highest PSNR and SSIM values. RetinexMamba's use of SS2D and Fused-Attention improves efficiency and interpretability, making it a promising solution for low-light image enhancement.RetinexMamba is a novel architecture combining Retinex theory with Mamba for low-light image enhancement. It addresses the limitations of traditional Retinex methods and deep learning approaches like Retinexformer by integrating the computational efficiency of State Space Models (SSMs) and introducing an innovative illumination estimator and damage restorer. RetinexMamba replaces the Illumination-Guided Multi-Head Attention (IG-MSA) in Retinexformer with a Fused-Attention mechanism, improving model interpretability. The architecture features an Illumination Estimator (IE) and an Illumination Fusion Visual Mamba (IFVM), with the IFVM using an Illumination Fusion State Space Model (IFSSM) to achieve linear computational efficiency. The IFSSM employs 2D Selective Scanning (SS2D) and an Illumination-Fused Attention (IFA) mechanism to enhance feature extraction and attention calculation. Experimental evaluations on the LOL dataset show that RetinexMamba outperforms existing deep learning approaches in both quantitative and qualitative metrics, demonstrating its effectiveness in enhancing low-light images. The model achieves higher PSNR and lower RMSE compared to state-of-the-art methods, while maintaining better image quality and reducing color distortion and noise. Ablation studies confirm the model's superiority, with the highest PSNR and SSIM values. RetinexMamba's use of SS2D and Fused-Attention improves efficiency and interpretability, making it a promising solution for low-light image enhancement.