The toxicology of metallic mercury on motor neurons and their processes requires further work to resolve controversial implications in the aetiology of human motor neuron disease (MND). The assessment of experimental neurotoxicity in the peripheral motor system is, however, technically problematic and difficult to interpret. The mean number of axons in a nerve can vary considerably due to a high degree of biological variation. Atrophy of large axons can appear as loss when, in fact, their numbers appear in smaller diameter axonal categories. We addressed these quantitative problems using the murine phrenic nerve (MPN), a mono-fascicular, predominantly motor nerve as a model system. One micrometer transverse sections of gluteraldehyde/osmium tetroxide fixed MPNs were stained for myelin using a silver technique. Axon areas were measured from digital images of the nerve in cross-section (ImagePro Plus software) and transformed to circular diameter equivalents, then displayed as frequency distributions. We found a high biological variation in the mean axon number between paired nerves within experimental groups. Therefore, axon diameter data within individuals group was pooled. Theoretical simulation of axonal degeneration, atrophy and hypertrophy of larger myelinated axons (also affected in MND) were modelled by manipulating the original data set. With this model, by comparing normal distributions, it is possible to distinguish axonal atrophy, degenerative loss, and hypertrophy as distinct pathological processes in the large calibre axon subgroup that are selectively vulnerable to metallic toxins such as mercury.