The GaussianVTON framework introduces a novel 3D Virtual Try-On (VTON) pipeline that leverages 3D Gaussian Splatting editing to enhance image-prompting 3D editing and 3D VTON. This method enables realistic try-on experiences by reconstructing and editing real scenes. To address challenges in transitioning from 2D to 3D editing, a three-stage refinement strategy is employed. Additionally, a specialized editing strategy called Edit Recall Reconstruction (ERR) enhances rendering smoothness and prevents undesirable artifacts resulting from complex geometry alterations. The framework focuses on direct image-prompting 3D editing, allowing for personalized editing of humans in 3D scenes compared to text-driven methods. The method showcases superior performance in various evaluation metrics and human evaluations, indicating precise editing and effective mitigation of image distortion.