We reviewed 355 nerve biopsies analysed at the Laboratories of Neuropathology of the Born-Bunge Foundation/University of Antwerp (BBF/UIA) and University of Liège (ULg) between 1987 and 1997. We examined the indications for nerve biopsy, the yield of the procedure, and the influence of clinical and neuropathological parameters. Contributory biopsies accounted for 35.5% and 47.3% respectively at ULg and BBF/UIA laboratories: of these, one third showed specific histological findings, the majority being informative only when combined with the relevant clinical data. The profile of indications for nerve biopsy was roughly comparable in both laboratories. The search for an inflammatory neuropathy prompted 35-40% of all biopsies with more than 50% of specimens being informative in this indication. The lowest yield (20%) was obtained among the nerve biopsies performed in the absence of any presumptive aetiology. These accounted for 22-33% of all cases. Inadequate surgical resection, delays in transport or processing errors precluded histological study of 4% (BBF/UIA) to 8% (ULg) of the specimens. We conclude that nerve biopsies should be performed by experienced surgeons and handled in specialised laboratories. Only a relatively small number of causes of neuropathy can be diagnosed on the basis of histology alone. More often, contributory biopsies will result from the combination of non-specific suggestive histological features with relevant clinical information. The diagnostic yield of nerve biopsy is function of careful patient selection and close collaboration between the clinician and the neuropathologist.