With the expansion of the market potential and product distribution of remote sensing data and images, the need of protection for those image products is getting more and more interest. Watermarking of remote sensing images faces to some specific requirements compared with general multimedia watermarking techniques, which relate to a very different range of researches and applications that those images are dedicated to. We propose a novel watermarking approach based on the feature of blocked discrete cosine transform to decrease the effect of watermarking on the application values of remotely sensed images. The algorithm is a blind one and watermark detection is possible both with the data in the JPEG compressed domain and in the spatial domain, thus increasing the flexibility and usability of the system. Experimental results show that the proposed watermarking system for remote sensing images is robust in the face of various image manipulations, including adding noise, cropping, filtering, and compression.