The total number of myelinated fibres in four unifascicular tibial nerves from diabetic rats has been counted and measured in order to assess the merits of various schemes for estimating group mean fibre diameter. The average nerve trunk contained some 2960 fibres which were measured in just under five hours. With different sampling designs, the average measurement time per nerve was reduced to between 17 and 69 minutes, with little consequent loss of reliability or precision of estimated mean fibre size. The most efficient schemes were those taking systematic samples of squares or sectors. A modification of a method relying on complete strips across two diameters of each nerve was the least efficient sampling approach. It had the additional disadvantage of introducing systematic errors which could affect the accuracy of measurements made on nerve trunks with heterogeneous spatial distributions of fibre size and number. This paper completes an investigation into random sampling errors influencing morphometric estimates of fibre size based on uni- or multifascicular nerve trunks.