The administration of an intravenous anaesthetic agent before experimental cerebral ischaemia in animals improves the functional and histological outcome. Cerebral ischaemia may be global or focal, complete or incomplete. Intravenous anaesthetic agents reduce the cerebral metabolic demand for oxygen (CMRO2) and abolish electrophysiological activity. This reflects a discontinuation of the functional neuronal activity with maintenance of its basic metabolic activity. The oxygen spared by the decrease in consumption, while reducing the functional activity, might be used by the neurons to sustain longer periods of ischaemia. This protective effect is also observed after pretreatment with either lidocaine or volatile agents, but their potentially deleterious vasodilating effect must be considered. Ketamine has recently been shown to antagonize NMDA receptors. The protective effect of barbiturates was experimentally demonstrated more than 30 years ago. They are still used as a reference. They reduce CMRO2, optimise the ratio between oxygen consumption and oxygen delivery and thus reduce cerebral blood flow and cerebral blood volume, as a result of the decrease of the metabolic demand. This might explain why a protective effect is seen in case of global or focal hypoxia with increased intracranial pressure, while no protection is documented in case of global cerebral ischaemia, such as after cardiac arrest, where EEG is immediately flat and ICP low. However, at doses required to obtain a protective effect, barbiturates induce deleterious side effects such as severe arterial hypotension, which limits their use. Cerebrovascular and cardiac surgery or surgery of the carotids are characterised by potentially ischaemic episodes which can be predicted.(ABSTRACT TRUNCATED AT 250 WORDS)