Human placentae from well-controlled diabetic women were collected after 37 weeks of gestation and divided into three groups according to the duration and severity of diabetes mellitus established by White classification criteria. A fourth group of subjects served as matched controls. Various morphometric variables not estimated hitherto (including the star volumes of villous 'domains' and intervillous 'pores' and trophoblast surface denudation) were assessed stereologically. The aims were to test whether or not (1) control values of these structural quantities are preserved in well-controlled diabetes mellitus, and (2) differences occurred between alternative diabetic groups. Placental specimens were obtained by systematic random sampling procedures and paraffin sections were cut at random positions and orientations. Volume densities of peripheral (terminal+intermediate) villi and intervillous spaces were estimated by test point counting and multiplied by placental volumes in order to convert them into absolute volumes. Volume estimates were also obtained for trophoblast, syncytiotrophoblast nuclei and intervillous fibrin-type fibrinoid. Villous surface areas were estimated by intersection counting and the star volumes of villi and intervillous pores were obtained by measuring the lengths of point-sampled intercepts. Calculations were also made of the theoretical numbers of villous domains and intervillous pores and of the numbers of syncytiotrophoblast nuclei. No significant differences were detected between control and diabetic placentae, or between White classes, for any of the estimated quantities. It is concluded that normal values are preserved by good glycaemic control regardless of diabetic grouping.