In this paper, we proposed a novel wavelet-tree based watermarking method by using structure-based quantization. First, we arrange wavelet-trees into super-trees. Secondly, the watermark bits are embedded into the super-trees by the proposed structure-based quantization method. According to these bits, the super-trees will be quantized into a significant structure. Compared to the unquantized super-tree, the quantized version has strong statistical character in energy distribution. Based on this character, the watermark bits could be extracted robustly from the attack of image distortion. The experimental results show that the proposed watermarking method performs well in JPEG compression and filtering (Gaussian filter, median filter and sharpen). In addition, it is also very robust to against the multiple watermark attack.