The focused ion beam-nanotomography (FIB-nt) technique presented in Part I of this article is a novel high-resolution three-dimensional (3D) microscopy method that opens new possibilities for the microstructural investigation of fine-grained granular materials. Specifically, FIB-nt data volumes allow particle size distributions (PSD) to be determined, and the current paper discusses all the processing steps required to obtain the PSD from 3D data. This includes particle recognition and the subsequent PSD estimation. A refined watershed approach for 3D particle recognition that tolerates concavities on the particle surfaces is presented. Particles at the edge of the 3D data volume are invariably clipped, and because the data volume is of a very limited size, this effect of boundary truncation seriously affects the PSD and needs to be corrected. Therefore, two basic approaches for the stereological correction of the truncation effects are proposed and validated on artificially modeled particle data. Finally, the suggested techniques are applied to real 3D-particle data from ordinary portland cement and the resulting PSDs compared with data from laser granulometry.