Ever since the seminal findings of Ramon y Cajal, dendritic and axonal morphology has been recognized as a defining feature of neuronal types and their connectivity. Yet our knowledge about the diversity of neuronal morphology, in particular its distant axonal projections, is still extremely limited. To systematically obtain single neuron full morphology on a brain-wide scale in mice, we established a pipeline that encompasses five major components: sparse labeling, whole-brain imaging, reconstruction, registration, and classification. We achieved sparse, robust and consistent fluorescent labeling of a wide range of neuronal types across the mouse brain in an efficient way by combining transgenic or viral Cre delivery with novel transgenic reporter lines, and generated a large set of high-resolution whole-brain fluorescent imaging datasets containing thousands of reconstructable neurons using the fluorescence micro-optical sectioning tomography (fMOST) system. We developed a set of software tools based on the visualization and analysis suite, Vaa3D, for large-volume image data processing and computation-assisted morphological reconstruction. In a proof-of-principle case, we reconstructed full morphologies of 96 neurons from the claustrum and cortex that belong to a single transcriptomically-defined neuronal subclass. We developed a data-driven clustering approach to classify them into multiple morphological and projection types, suggesting that these neurons work in a targeted and coordinated manner to process cortical information. Imaging data and the new computational reconstruction tools are publicly available to enable community-based efforts towards large-scale full morphology reconstruction of neurons throughout the entire mouse brain.