Three methods of classifying laboratories during fibre counting proficiency tests were compared. The first two are those used in France (classification according to the mean and coefficient of variation of the results) and in Great Britain (classification according to the proportion of normalized results situated within predefined limits). The third is a variation of the Workplace Analysis Scheme for Proficiency (WASP) programme adapted to fibre counting tests. In the latter case, the laboratory classification is based on comparing the variance characterizing the dispersion of the results of a laboratory with a reference variance, which is considered as the variance of experienced analysts or laboratories. This mode of processing has the advantage of allowing the comparison of magnitudes. For example, the variance of the reference value can be compared with the reference variance. The same applies if a proficiency test is organized on the basis of replicas distributed to different analysts, the variability of these replicas can be compared with the reference variance. It emerged that the modified WASP method produces results close to those obtained by the other two methods. Moreover, the selectivity of the three methods is evaluated.