In two groups of anaesthetized dogs, with (n = 28) or without (n = 28) induced intracranial hypertension, we compared the effects on intracranial pressure (ICP) of the rapid administration of mannitol 2 g kg-1 i.v. at PaCO2 2.7, 4.0, 5.3, and 6.7 kPa (n = 7). In dogs with no induced intracranial hypertension, ICP increased during the administration of mannitol, reached a peak at 2 min after infusion, and then gradually decreased (P less than 0.05). More marked changes in ICP were observed in response to higher values of PaCO2 (P less than 0.05). In dogs with induced intracranial hypertension, the rapid infusion of mannitol caused an exponential decrease in ICP, without initial increase, which was significantly steeper at higher values of PaCO2 (P less than 0.05). This was followed by a more gradual decrease which achieved pre-balloon inflation values 10 min after infusion. We postulate that the absence of the initial increase in ICP is the result of a concomitant decrease in arterial pressure, a reduction in the volume-pressure response of the brain, the failure of mannitol to dilate further the cerebral arterial vascular bed and a hitherto unnoticed early water-drawing effect. Our study confirmed the safety of rapidly expanding the circulating blood volume with mannitol in circumstances of increased ICP in dogs.