In recent years, there have been important advances in the quantification of high-resolution (electron microscopical) images of tissue sections on which colloidal gold-labelled probes serve to identify and localize interesting target antigens. With these new methods, the distributions of gold particle counts across volume-occupying and/or surface-occupying compartments can be compared within or between experimental groups of cells, tissues or organs. Method I (for within-group comparisons) tests whether or not there is preferential labelling of compartments by comparing observed and expected gold labelling distributions using Chi-squared (chi2) analyses. To this end, estimators of gold labelling intensity (labelling density, LD, and/or relative labelling index, RLI) are used to analyse one category of compartment (volume or surface occupiers) or a mixture of categories (volume and surface occupiers). This involves estimating compartment size taking into account specimen magnification and, in its most efficient form, is achieved simply by counting chance events after superimposing random test probes (lattices of points and/or lines). RLI=1 indicates random labelling but higher RLI values indicate the degree to which a compartment departs from random labelling. Method II (drawing between-group comparisons) does not require information about compartment size or specimen magnification. Comparisons are drawn using the observed gold particle counts in different groups and by combining chi2 analysis with contingency table analysis. Both methods rely on multistage systematic uniform random sampling of specimens and unbiased counting of gold particles associated with different compartments. Statistical degrees of freedom are determined by the numbers of compartments and experimental groups. Together with RLI values, compartmental values of chi2 which contribute substantially to total chi2 identify the principal sites of within- and between-group differences. The methods are illustrated using data taken from studies aimed at localising protein antigens (caveolin-1 and GLUT1) in specimens of term human placenta.