Although diffusion models can generate high-quality human images, their applications are limited by the instability in generating hands with correct structures. In this paper, we propose a conditional diffusion-based framework, RHanDS, to refine the malformed hand with the help of decoupled structure and style guidance.The structure guidance is the hand mesh reconstructed from the malformed hand, serving to correct the hand structure. The style guidance is the malformed hand itself and is employed to furnish the style reference for hand refining.To alleviate the mutual interference between style and structure guidance, we introduce a two-stage training strategy and build a series of multi-style hand datasets. In the first stage, we use paired hand images for training to generate hands with the same style as the reference. In the second stage, various hand images generated based on the human mesh are used for training, enabling the model to gain control over the hand structure while maintaining style consistency.Experimental results show that RHanDS can effectively refine hand structure while maintaining consistency in hand style. The code and dataset will be available once this paper is accepted.