The mechanisms which lead to secondary brain damage following transient ischaemia are incompletely defined. As discussed in this hypothesis article, the events which lead to such damage could encompass (a) a perturbed membrane handling of calcium, leading to a slow, gradual increase in the free cytosolic calcium concentration (Ca2+i), with subsequent calcium overload of mitochondria, (b) a sustained reduction of protein synthesis which, in the long run, deprives cells of enzymes or trophic factors essential to their survival, or (c) the initiation of an inherent program for cell death. Results obtained in ischaemia of brief to intermediate duration demonstrate that the ultimate cell death is heralded by a reduction in the respiratory capacity of isolated mitochondria. However, the results fail to demonstrate whether or not such a reduction precedes deterioration of the bioenergetic state which then precipitates cell death. Cyclosporin A (CsA) has recently been shown to dramatically improve the delayed CA1 damage following transient forebrain ischaemia. Since CsA is known to block a deleterious permeability transition (PT) in mitochondria from several tissues in response to calcium accumulation and oxidative stress, the results on CsA effects in forebrain ischaemia support a mitochondrial origin for the delayed cell death. Furthermore, comparisons with the effects of CsA and α-phenyl-N-tert-butyl nitrone (PBN) in thymocytes and other cells undergoing programmed cell death suggest that delayed neuronal damage occurs by a sequence of events akin to those leading to apoptotic cell death. However, whether cell death is apoptotic or necrotic may depend on the severity of the insult (and its duration). We speculate that the initial ischaemic transient leads to gradual mitochondrial calcium overload, the latter triggering a PT, and apoptotic or necrotic cell death. Since similar results have been obtained in normoglycaemic animals subjected to ischaemia of intermediate duration, and in animals with preischaemic hyperglycaemia, it seems likely that both increased ischaemia duration and hyperglycaemia accelerate damage to mitochondria in the reperfusion period. Recent results obtained in transient focal ischaemia of 2 h duration demonstrate that the free radical spin trap PBN reduces infarct size, even when given 1 or 3 h after the start of reperfusion, thus providing a second window of therapeutic possibility. A major effect of the drug is exerted on the recovery of energy metabolism of the tissue since it reduces a secondary deterioration in the bioenergetic state, occurring after 2–4 h of reperfusion. At least in part, the spin trap may exert its effect by reducing microvascular dysfunction caused by oedema and to adhesion of polymorphonuclear (PMN) leucocytes, which give rise to an inflammatory response mediated by cytokines, lipid mediators, or free radicals. This contention is supported by the reduction in focal ischaemic damage by antibodies to adhesion molecules for PMNs. However, it has now been found that the secondary deterioration of the bioenergetic state of core and penumbral tissues are mirrored by corresponding changes in the respiratory functions of isolated mitochondria, suggesting that, also in this type of ischaemia, the mitochondria suffer secondary damage. It is conceivable that a significant fraction of malfunctioning mitochondria emanate from microvascular tissue, explaining why antibodies to adhesion molecules mitigate the ischaemic lesions.