A single biopsy is a poor estimator of the muscle fibre cross-sectional area (CSA) for a whole human muscle because of the large variability in the fibre area within a muscle. To determine how the sampling errors in biopsy techniques can be reduced, data on the CSA of type 1 and type 2 fibres obtained from cross-sections of whole vastus lateralis muscle of young men, have been analysed statistically. To obtain a good estimate of the mean fibre CSA in a biopsy, measuring all fibres in that biopsy gives the best result. To obtain a good estimate of the mean fibre CSA for a whole muscle, the number of biopsies has a much greater influence on the sampling error than the number of fibres measured in each biopsy, but the number of biopsies needed to obtain a given sampling error can vary by a factor of two. If the fibre CSA in three or more biopsies is measured, it is sufficient to measure only 25 fibres in each biopsy. If less than three biopsies are taken, there is no worthwhile reduction in sampling error when more than 100 fibres are measured. To determine the mean fibre CSA for a whole group of individuals, our preference is to maximize the number of individuals, and only take single biopsies. In conclusion, to determine the mean fibre CSA for this particular muscle with a certain precision, we suggest analysis of three biopsies, taken from different depths of the muscle, and measurement of 25 fibres in each biopsy.