This paper presents a novel interactive 3D axon tracking and labeling tool, AXONTRACKER-3D, designed to reconstruct axonal structures in the neuromuscular junction connectome. The tool enables the extraction of quantitative information from volumetric axon imagery by automatically tracking and labeling axons in 3D. The software is freely available for download and is designed to be user-friendly and accurate, making it an attractive tool for neurobiological studies. The neuromuscular junction connectome is a promising starting point for studying the organization and development of the mammalian central nervous system. It is well-defined and accessible, with large axons separated by Schwann cells and large neuromuscular junctions, making it easier to resolve and analyze. The morphological features of the neuromuscular connectome are directly linked to its functional properties, making it an attractive platform for connectome studies. The paper describes the development of AXONTRACKER-3D, which incorporates a 3D, real-time tracking approach based on diffusion. This approach significantly reduces the time required to process large volumes of axon images from months to minutes on personal computers. The tool is designed to handle the complexity of axon structures, including their spatial organization, branching patterns, and geometric information. The algorithm used in AXONTRACKER-3D involves anisotropic diffusion image filtering to reduce noise while preserving contrast near the edges of the axons. Centerline extraction is performed using a combination of gradient vector flow (GVF) and object orientation vectors. The centerlines are then used for segmentation and visualization. The software also includes a provision for manual intervention in cases where the algorithm fails to find the correct center due to poor contrast or low intensity. The paper also discusses the challenges in segmenting axons, including weak boundaries between close-lying axons and large intensity variations. To address these challenges, the segmentation scheme consists of two steps: local threshold 3D region growing and interactive level set modeling. The datasets are then stitched together to form a collage, which is used to visualize the connectome. The study highlights the importance of quantitative data from tracking and segmentation of complete connectomes in generating biological questions about the relationship between the number of branches and the length of each segment in a population of axons, as well as the spatial distribution of motor neurons. The paper also discusses planned improvements to the software, including the detection and tracking of branches in axons to minimize user intervention.This paper presents a novel interactive 3D axon tracking and labeling tool, AXONTRACKER-3D, designed to reconstruct axonal structures in the neuromuscular junction connectome. The tool enables the extraction of quantitative information from volumetric axon imagery by automatically tracking and labeling axons in 3D. The software is freely available for download and is designed to be user-friendly and accurate, making it an attractive tool for neurobiological studies. The neuromuscular junction connectome is a promising starting point for studying the organization and development of the mammalian central nervous system. It is well-defined and accessible, with large axons separated by Schwann cells and large neuromuscular junctions, making it easier to resolve and analyze. The morphological features of the neuromuscular connectome are directly linked to its functional properties, making it an attractive platform for connectome studies. The paper describes the development of AXONTRACKER-3D, which incorporates a 3D, real-time tracking approach based on diffusion. This approach significantly reduces the time required to process large volumes of axon images from months to minutes on personal computers. The tool is designed to handle the complexity of axon structures, including their spatial organization, branching patterns, and geometric information. The algorithm used in AXONTRACKER-3D involves anisotropic diffusion image filtering to reduce noise while preserving contrast near the edges of the axons. Centerline extraction is performed using a combination of gradient vector flow (GVF) and object orientation vectors. The centerlines are then used for segmentation and visualization. The software also includes a provision for manual intervention in cases where the algorithm fails to find the correct center due to poor contrast or low intensity. The paper also discusses the challenges in segmenting axons, including weak boundaries between close-lying axons and large intensity variations. To address these challenges, the segmentation scheme consists of two steps: local threshold 3D region growing and interactive level set modeling. The datasets are then stitched together to form a collage, which is used to visualize the connectome. The study highlights the importance of quantitative data from tracking and segmentation of complete connectomes in generating biological questions about the relationship between the number of branches and the length of each segment in a population of axons, as well as the spatial distribution of motor neurons. The paper also discusses planned improvements to the software, including the detection and tracking of branches in axons to minimize user intervention.