Detecting, localising and counting ultrasmall particles and nanoparticles in sub- and supra-cellular compartments are of considerable current interest in basic and applied research in biomedicine, bioscience and environmental science. For particles with sufficient contrast (e.g. colloidal gold, ferritin, heavy metal-based nanoparticles), visualization requires the high resolutions achievable by transmission electron microscopy (TEM). Moreover, if particles can be counted, their spatial distributions can be subjected to statistical evaluation. Whatever the level of structural organisation, particle distributions can be compared between different compartments within a given structure (cell, tissue and organ) or between different sets of structures (in, say, control and experimental groups). Here, a portfolio of stereology-based methods for drawing such comparisons is presented. We recognise two main scenarios: (1) section surface localisation, in which particles, exemplified by antibody-conjugated colloidal gold particles or quantum dots, are distributed at the section surface during post-embedding immunolabelling, and (2) section volume localisation (or full section penetration), in which particles are contained within the cell or tissue prior to TEM fixation and embedding procedures. Whatever the study aim or hypothesis, the methods for quantifying particles rely on the same basic principles: (i) unbiased selection of specimens by multistage random sampling, (ii) unbiased estimation of particle number and compartment size using stereological test probes (points, lines, areas and volumes), and (iii) statistical testing of an appropriate null hypothesis. To compare different groups of cells or organs, a simple and efficient approach is to compare the observed distributions of raw particle counts by a combined contingency table and chi-squared analysis. Compartmental chi-squared values making substantial contributions to total chi-squared values help identify where the main differences between distributions reside. Distributions between compartments in, say, a given cell type, can be compared using a relative labelling index (RLI) or relative deposition index (RDI) combined with a chi-squared analysis to test whether or not particles preferentially locate in certain compartments. This approach is ideally suited to analysing particles located in volume-occupying compartments (organelles or tissue spaces) or surface-occupying compartments (membranes) and expected distributions can be generated by the stereological devices of point, intersection and particle counting. Labelling efficiencies (number of gold particles per antigen molecule) in immunocytochemical studies can be determined if suitable calibration methods (e.g. biochemical assays of golds per membrane surface or per cell) are available. In addition to relative quantification for between-group and between-compartment comparisons, stereological methods also permit absolute quantification, e.g. total volumes, surfaces and numbers of structures per cell. Here, the utility, limitations and recent applications of these methods are reviewed.