In this paper, we propose a novel method for secure tamper localization in binary document images using erasable watermarks. For binary images, watermarking with pixel flipping approach is a difficult task, because it can bring noticeable visual distortion. In localization, finding sufficient number of low-distortion pixels in a block to embed the cryptographic signature and their blind detection is more difficult. Also, an imperceptible watermark cannot be embedded in white regions of the document image, making such regions insecure against hostile attacks. In the proposed new method, an erasable watermark is embedded in each block of a document image independently for secure localization. The embedding process introduces some background noise; however the content in the document can be read or understood by the user, because human vision has the inherent capability to recognize various patterns in the presence of noise. After verifying the content of each block, the exact copy of original image can be restored at the blind detector for further analysis. In the proposed method, the tamper localization accuracy is significantly improved as compared to the method proposed by Kim and Queiroz. Simulation results show that an erasable watermark of necessary data length can be embedded in individual blocks of various document images to attain cryptographic security.