Thinner powder layers are beneficial to three-dimensional printing (3DP) parts in many aspects, such as accuracy, surface quality and densification. However, counter-rolling (CR) layering (belonging to dry-powder layering method) and slurry-based layering, which are both conventional layering methods, either cannot achieve ultra-thin layering due to the considerable layering defects (including the cavity defect and part-shifting defect) or must involve complicated processes. Therefore, to keep the convenience of dry-powder layering and avoid the excessive layering defects, a feasibility study of the double-smoothing (DS) method for ultra-thin layering was conducted on a self-developed 3DP machine in this work. Furthermore, with optical monitoring, layering defects were first investigated experimentally. It is proved that DS layering method is capable of dispensing dry powder into intact ultra-thin (55μm) layers: the cavity defects were well restricted, the layer-location deviations of the printed green parts were within 100μm per 218 layers, the green densities exceeded 70%, and the uniform structures within the printed specimens were also realized. Combining a modified Mohr–Coulomb failure theory (Jenike yield theory) and the solution of in-powder stress fields induced during powder layering, a theoretical framework was established to primarily interpret the restriction of layering defects benefiting from the DS layering, whereby this theoretical framework as a reference tool for future 3DP design works was also implied.