Patients with acute occlusion of a major cerebral artery sequentially call on a hemodynamic reserve quantifiable in terms of rCBF and rCBV and then an oxygen carriage reserve quantifiable in terms of rOER (Fig. 10). If both are exhausted, ischemia supervenes and rCMRO2 becomes linearly related to rCBF. Depending on the degree and duration of fall in rCMRO2, variable degrees of infarction occur. Infarction occurs more rapidly in subcortical than cortical tissue. The phase of maximal rOER is short, and rapid evolution by either reflow or delayed cell death results in an invariable decline of rOER to subnormal levels in the face of a low rCMRO2 (Fig. 11). This combination is characteristic of infarcted tissue. In some instances, partial infarction with a low perfusion reserve is observed with low rCMRO2, high though submaximal rOER, and no increase in rCMRO2 if rCBF is raised. This is a precarious situation predisposing to extension of infarction and suggests blood pressure should not be lowered acutely after a stroke unless the latter is caused by hypertensive encephalopathy. This pattern, however, is rare, short-lived, and usually followed by late reflow or further infarction as a further fall in rCPP occurs, transiently or permanently. Tests of rCPP and the two homeostatic mechanisms involved in maintaining rCMRO2 using non-PET techniques suggest themselves and might form an objective basis for selection of patients for prophylactic revascularization or bypass operations. PET techniques are now becoming available for assessing the blood-brain barrier, tissue pH, and amino acid metabolism, all of which may have relevance to the further understanding of the evolution of infarcts. The rapid study of very early ischemic events is now required to elucidate further the potential for cerebral salvage therapy once ischemia has occurred.