The aim of the study was to evaluate the precision of the nucleator principle with regard to estimation of follicle, oocyte and antral volumes in individual follicles, as well as their average volumes in a group of follicles. Nine isolated murine ovarian follicles embedded in plastic were orientated isotropically, cut into 20 microns serial sections, and stained with hematoxylin. In each follicle the volumes of the follicle, the antrum and the oocyte were estimated in two ways, both unbiased: Firstly, in just one unique section through the follicle, viz. the section which contains the unique sampling point--the nucleolus of the oocyte--using the nucleator principle, and, secondly, in the complete set of follicle sections using the Cavalieri principle. Estimates using the Cavalieri, which are very precise (CEFollicle approximately 0.03, CEOocyte approximately 0.11, CEAntrum approximately 0.16), were interpreted as the "true" values. Therefore, the precision (estimator variation) of the nucleator estimates could be calculated. The nucleator estimates correlated well with the Cavalieri estimates: r(follicle) = 0.95; r(antrum) = 0.88; r(oocyte) = 0.83. The nucleator provides very precise estimates of the volumes of follicle, oocyte and antrum with a minimal amount of effort, since the estimates are derived from measurements in only four directions sampled in a uniform systematic way, the first direction being chosen at random, in only one section through the follicle. Moreover, it is not necessary to know the precise thickness of the section, and finally, the estimates are expressed in correctly dimensioned terms. The estimator variation of the nucleator (CE = 0.28) is of minor importance compared to the biological variation (CV = 0.47) in this realistic sample of follicles.