Mapping from optical satellite images without ground control points (GCPs) is one of the goals of photogrammetry. Aiming at the mapping without GCPs, a large-scale block adjustment method based on the virtual control points is proposed in this paper, in which the single image is regarded as an adjustment unit to be organized. To overcome the block distortion caused by unstable adjustment without GCPs and the excessive accumulation of errors, the virtual control points created by the initial RPC model of the images are regarded as the weighted observations and introduced into the adjustment model to refine the adjustment. To verify the availability and the accuracy of the presented method, 26 406 three-linear-array images of ZY-3 covering China are used as the basic data and 8000 uniformly distributed high precision check points are applied to evaluate the geometric accuracy of the DOM (digital ortho model) and DSM (digital surface model) production. The experimental results indicate that the standard deviations of plane and elevation are both better than 4 m and the mosaic accuracy of neighboring DOM is within better than 1 pixel, which could meet the demand of seamless mosaic.