Traditional spaceborne Synthetic Aperture Radar (SAR) can acquire three-dimensional (3D) images by using multi-baseline SAR data, which solves the problem of layover in two-dimensional (2D) SAR image, however, there is still a problem of insufficient information acquisition caused by occlusion. Considering this, a 3D imaging algorithm for multi-azimuth-angle multi-baseline spaceborne SAR is proposed in this paper; this algorithm not only solves the problem of layover, but also reduces the occlusion area by fusing the 3D point clouds obtained under different azimuth angles and improves the information acquisition ability of spaceborne SAR. First, the observation model of multi-azimuth-angle multi-baseline spaceborne SAR is established, and it shows that the mathematical signal models of the multi-baseline SAR in the squint mode and broadside mode are the same, which provides theoretical support for the direct application of broadside 3D imaging algorithm to the 3D processing of squint mode. On this basis, the 3D imaging algorithm and processing flow of multiazimuth-angle multi-baseline spaceborne SAR are presented. Finally, the effectivenesses of the squint 3D imaging method and 3D point clouds fusion method are verified by point targets simulation experiment with 45° azimuth angle and helicopter simulation experiments with 45° and –45° azimuth angles, respectively.