Biopsies were made on six rat tibial nerves and 'absolute' numbers of myelinated fibres determined by counting all fibres in all nerve trunks. Subsequently, two unbiased sampling schemes-systematic random quadrat (SRQ) and 'fractionator' sampling-were used to select fibres from the same nerves and to obtain estimates of their numbers. Both schemes captured roughly 200 out of the total of 3000 myelinated fibres found in these nerve trunks, all of which were sampled exhaustively, without replacement. Estimates of fibre number were derived by an established approach, the 'ratio technique' (using SRQ samples), and by a new principle, the fractionator (using fractionator samples). Counting all fibres in every nerve trunk took almost 6 hours. Both the ratio technique and the fractionator approaches provided efficient and unbiased estimates of fibre numbers. Six nerve trunks were analysed by SRQ sampling in 77 minutes, compared with 65 minutes by the fractionator. Apparent differences between the two approaches were of minor interest when set against the benefits of sampling per se. These findings are likely to be of practical concern to those wishing to examine nerves with great numbers of fibres and/or to examine large numbers of nerves.