We describe a method for unbiased assumption-free estimation of the total number of beta-cells and the mean beta-cell volume in mouse and rat pancreas based on light microscopy. Such a method, which takes advantage of one of the most recent developments in stereology, the fractionator, has not previously been described. It relies on repeated fractionation of the tissue using systematic uniform random sampling combined with an unbiased counting principle. The method was applied to eight BALB/cBom male mice (56 days) and six Lewis/MOL male rats (47 days). In mice, the total number of beta-cells was 1.06 +/- 0.07 x 10(6) (mean +/- SEM) per pancreas with a mean beta-cell volume of 1280 +/- 17 microm3, while in rats the total beta-cell number was 2.76 +/- 0.42 x 10(6) per pancreas with a mean beta-cell volume of 1170 +/- 65 microm3. Furthermore, the results showed that in both species the biological variability in the total beta-cell volume is due to differences in the number of beta-cells rather than variability of the mean beta-cell volume. The method can be used to give a precise description of number and volume of beta-cells at different ages, and will make it possible to estimate the contributions of hyper/hypotrophia and hyper/hypoplasia to a given induced or spontaneous change in the total beta-cell mass.