The paper introduces a new algorithm called *marching cubes* for creating triangle models of constant density surfaces from 3D medical data, such as CT, MRI, and SPECT images. The algorithm uses a divide-and-conquer approach to generate inter-slice connectivity, creating a case table that defines triangle topology. It processes the 3D data in scan-line order, calculates triangle vertices using linear interpolation, and determines the gradient of the original data to shade the models. The detail in the images produced is attributed to maintaining inter-slice connectivity, surface data, and gradient information. The paper discusses improvements that reduce processing time and add solid modeling capabilities, and presents case studies using CT, MRI, and SPECT data to illustrate the algorithm's quality and functionality. The marching cubes algorithm is implemented in C and can run on various platforms, with execution times depending on the number of surfaces and resolution of the original data. The results show that the algorithm produces high-resolution, detailed surface models that complement 2D medical images, providing physicians with 3D views of the anatomy.The paper introduces a new algorithm called *marching cubes* for creating triangle models of constant density surfaces from 3D medical data, such as CT, MRI, and SPECT images. The algorithm uses a divide-and-conquer approach to generate inter-slice connectivity, creating a case table that defines triangle topology. It processes the 3D data in scan-line order, calculates triangle vertices using linear interpolation, and determines the gradient of the original data to shade the models. The detail in the images produced is attributed to maintaining inter-slice connectivity, surface data, and gradient information. The paper discusses improvements that reduce processing time and add solid modeling capabilities, and presents case studies using CT, MRI, and SPECT data to illustrate the algorithm's quality and functionality. The marching cubes algorithm is implemented in C and can run on various platforms, with execution times depending on the number of surfaces and resolution of the original data. The results show that the algorithm produces high-resolution, detailed surface models that complement 2D medical images, providing physicians with 3D views of the anatomy.