In the present study we have used the detection of Fos, the protein product of c-fos, to determine the distribution of neurons in the medulla and hypothalamus that are activated by changes in central blood volume. Experiments were conducted in both barointact and barodenervated conscious rabbits, to determine the contribution of arterial baroreceptors to the pattern of Fos expression evoked by changes in central blood volume, induced either by intravenous infusion of an isotonic modified gelatin solution, or by partial occlusion of the vena cava. These procedures resulted in a significant increase and decrease, respectively, in right atrial pressure over a 60 min period. In control experiments, barointact and barodenervated rabbits were subjected to the identical procedures except that no changes in central blood volume were induced. In comparison with the control observations, central hypervolaemia produced a significant increase in the number of Fos-immunoreactive neurons in the nucleus tractus solitarius, area postrema, the caudal, intermediate and rostral parts of the ventrolateral medulla, supraoptic nucleus, paraventricular nucleus, arcuate nucleus, suprachiasmatic nucleus and median preoptic nucleus. The overall pattern of Fos expression induced by central hypervolaemia did not differ significantly between barointact and barodenervated animals. Similarly, the overall pattern of Fos expression induced by central hypovolaemia did not differ significantly between barointact and barodenervated animals, but did differ significantly from that produced by hypervolaemia. In particular, central hypovolaemia produced a significant increase in Fos expression in the same regions as above, but also in the subfornical organ and organum vasculosum lamina terminalis. In addition, compared with central hypervolaemia, hypovolaemia produced a significantly greater degree of Fos expression in the rostral ventrolateral medulla and supraoptic nucleus. Furthermore, double-labelling for tyrosine hydroxylase immunoreactivity demonstrated that neurons in the ventrolateral medulla that expressed Fos following hypovolaemia were predominantly catecholamine cells, whereas following hypervolaemia they were predominantly non-catecholamine cells. Finally, double-labelling for vasopressin immunoreactivity demonstrated that the number of Fos/vasopressin immunoreactive cells in the supraoptic nucleus was approximately 10 times greater following hypovolaemia compared with hypervolaemia, but there were very few such double-labelled neurons in the paraventricular nucleus in response to either stimulus. The results demonstrate that central hypervolaemia and hypovolaemia each induces reproducible and specific patterns of Fos expression in the medulla and hypothalamus. The degree and pattern of Fos expression was unaffected by arterial baroreceptor denervation, indicating that it is primarily a consequence of inputs from cardiac receptors, together with an increase in the level of circulating hormones such as atrial natriuretic peptide, angiotensin II or vasopressin. Furthermore, the pattern of Fos expression produced by central hypervolaemia and hypovolaemia is distinctly different from that evoked by hypertension and hypotension, respectively [Li and Dampney (1994) Neuroscience 61, 613-634], particularly in hypothalamic regions. These findings therefore indicate that the central pathways activated by changes in blood volume are, at least in part, separate from those activated by changes in arterial pressure.