We estimated the total number of calcitonin-immunoreactive C-cells in rat thyroid gland using the optical fractionator, the unbiased stereological method for estimation of numbers. It was necessary first to use a fixative composed of formalin, acetic acid, and ethanol to distinctly visualize the C-cells. The 40-microm-thick sections had to adhere to chromalum-gelatin-coated Superfrost Plus glass slides, and the immunostaining technique had to stain the C-cells evenly throughout the whole sections. Because the C-cells were irregularly distributed in the thyroid tissues, their counting required screening of about 500 fields per lobe, but the number of C-cells counted need not be high, about 90 per lobe. We estimated that rats have 185,000 +/- 42,000 C-cells (mean +/- SD; n - 7). The C-cell population did not differ significantly between the two lobes of a given rat, but it varied markedly among rats. The biological differences among the animals contributed 83% to the observed variability, whereas the methodological uncertainty contributed 17%. The serum levels of calcitonin and calcium were not closely correlated to the C-cell numbers. Our results indicate that variability in C-cell experiments can be reduced most effectively by increasing the number of animals used. However, the similar C-cell frequency found in the two thyroid lobes of each rat allows the use of one uniformly sampled lobe for quantification and the other lobe for further analysis.