We have compared asbestos fibre and general fibre counts from scanning electron microscopy (SEM) and transmission electron microscopy (TEM) with fibre counts from phase-contrast optical microscope (PCOM) methods. Three different types of sample have been evaluated: laboratory-prepared samples of different fibre types; chrysotile asbestos textile factory samples; and non-occupational and environmental samples from sites where asbestos might be found. TEM produced total fibre number assessments which were greater than those found with SEM which, in turn, produced fibre counts greater than those obtained with the PCOM. However, when fibres longer than 5 microns were alone counted, the two electron microscope (EM) methods provided similar results. This indicates that TEM is advantageous in comparison with SEM when counting or sizing short fibres and has no advantage for fibres longer than 5 microns. For fibres longer than 5 microns in both the laboratory-prepared and asbestos factory samples, the EM asbestos counts were higher than the PCOM fibre counts, the ratios depending on the fibre type in the case of the former. The PCOM fibre counts in samples from the non-occupational situations were shown to be poor predictors of airborne asbestos fibre concentrations determined by EM. This was mainly due to the presence of high and variable proportions of non-asbestos fibres in these samples. It is concluded that, in order to convert EM asbestos fibre concentrations (greater than 5 microns) to equivalent PCOM asbestos concentrations, they should be divided by 4.0 for chrysotile and 1.7 for amphibole asbestos.