A zero-watermarking scheme for color images to resist printing/scanning attack is designed in order to improve watermark robustness to resist printing/scanning attack. After the host images are layered, discomposed by Haar wavelet once and partitioned, the maximum singular value of each block is obtained as an image feature to construct zero-watermarking information. The detection process of watermarking is to layer the printed images, to adjust their brightness, to discompose them by Haar wavelet once, to partition them, and then to extract the maximum singular value of each block as features of printed images to construct zero-watermarking information of the printing/scanning image. The similarities of zero-watermark information before and after printing/scanning are compared. If identical, it indicates that the scheme could resist printing/scanning attack. As the experiment results show, zero watermark information of the host image before and after printing/scanning is identical which indicates the scheme has the ideal performance in resisting printing/scanning attack.