Abstract This paper presents a new Planar Least-Squares Projection (PLSP) method to accurately construct the layered contours directly from the point cloud for additive manufacturing. With the rapid development of 3D measuring technology, the dense and accurate point cloud, which can represent more accurate geometry information of a physical object than before, has been readily available. Additive manufacturing based directly on such point cloud is considered to be a promising alternative for fabricating the complex parts, but it requires the research on effective processing methods of the point set surface. In this paper, the presented PLSP method addresses this technical challenge for additive manufacturing. Mathematical modeling and processing of point cloud are presented to identify projections and construct accurate layered contours by considering both the least-square projection errors and projection distance selection criteria from measured point cloud. Technique of assigning new set of weights on measure scattered point cloud is discussed to achieve accurate layered contour development. The presented PLSP method with the new weight function and the consideration of point cloud deviation distances in sliced contour development is able to avoid the occurrence of the wrong projection that many other current methods suffer from. Technique of integrating the presented PLSP with the boundary point sequence curve (PSC) method is also discussed in this paper to eliminate contour development errors by considering the similarity in shape and closeness in distance for the ideal layered contours. Since this presented method does not involve the non- linear optimization, it is mathematically robust, and it can ensure the constructed layered contours lie accurately on the underlying nominal surface, and also has the ability of recognizing and processing the multi-contours on a common slicing plane. Both computational and experimental results based on the presented method are presented for validation.