Morphometric studies of the peripheral nerve often present widely varying results which, in part at least, may be attributed to the different methodologies used. Two questions may be of importance with regard to the reliability of such results: the preselection of fibres according to their morphology and the method used in quantifying observations. In this work a morphological study was carried out on the myelinated fibres of the sciatic nerve of a rat in order to evaluate fibre selection criteria. A morphometric analysis was also performed using manual measurements, image digitalisation and surveying, and automatic image analysis. It was shown that morphological variability of transverse section fibres is considerable and that, really, the proportion of circular fibres with homogeneous compact myelin is only 50 to 70%, from which we can conclude that the selection of fibres carried out in some studies wishing to eliminate abnormal fibres is somewhat exaggerated. By analysing the fibres using various methods significant differences appear, some due to the fact that distinct methods may be used to calculate the same parameters in a different way. The most reliable parameters would appear to be those which do not depend on the shape of the fibres and those which automatic or semiautomatic methods can calculate directly such as areas and perimeters. In any case quantifying methods seem hardly discriminatory and the differences between methods disappear if analyses are carried out using random samples. Preselection of fibres appears unnecessary in this context as in no case are the results altered. Finally we suggest finishing quantitative analyses with qualitative studies which would permit getting more information especially useful in cases of ageing, regeneration or pathological studies.