Perivascular nerve fibres of the uterine artery of virgin and late pregnant guinea-pigs were examined under the electron microscope following loading with 5-hydroxydopamine, a marker for catecholamine uptake, and immunohistochemistry for dopamine beta hydroxylase, neuropeptide Y, vasoactive intestinal polypeptide, substance P and calcitonin gene-related peptide. Varicosities, loaded with 5-hydroxydopamine labelled vesicles, and immunoreactive axons were counted in whole transverse sections of uterine arteries. Localization of the immunoreactivities in 5-hydroxydopamine-labelled vesicles was also studied. Colocalization of substance P and dopamine beta hydroxylase immunoreactivities was investigated at the light microscopic level. Both total and relative number of varicosities with 5-hydroxydopamine-labelled vesicles in a whole section of the artery increased in late pregnancy (61.2 +/- 10.2 versus 24.5 +/- 3.2 in virgin, representing 35% and 27% respectively, of all varicosities). Also the number of neuropeptide Y, vasoactive intestinal polypeptide, substance P and calcitonin gene-related peptide-immunoreactive axons increased, but their relative proportion remained unchanged. In virgin guinea-pigs only calcitonin gene-related peptide and neuropeptide Y immunoreactivities were associated with varicosities loaded with small dense-cored vesicles, while in late pregnancy 5-hydroxydopamine-labelled vesicles were also seen in a number of vasoactive intestinal polypeptide, substance P and calcitonin gene-related peptide-immunoreactive axons. Double immunolabelling for dopamine beta hydroxylase and substance P immunoreactivity showed that substance P immunoreactivity was not present in dopamine beta hydroxylase-immunoreactive axons of the uterine artery, of neither virgin nor late pregnant guinea-pigs. It is concluded that vascular hypertrophy of the uterine artery in late pregnancy is associated with an increase in the number of perivascular nerve fibres, that involves many, if not all of the subpopulations of neurons supplying the uterine artery. Also 5-hydroxydopamine-labelled varicosities were increased, but the results of the present study indicate that some of the nerve fibres that are able to take up 5-hydroxydopamine in late pregnancy are not sympathetic (i.e. are sensory and/or parasympathetic in origin). The relevance of these findings in pregnancy is discussed.