The ischaemic penumbra has been documented in the laboratory animal as a severely hypoperfused, non-functional, but still viable cortex surrounding the irreversibly damaged ischaemic core; with elapsing time, more penumbra gets recruited into the core, while tissue reperfusion is able to stop this deleterious process until a certain point in time. As saving the penumbra should improve clinical outcome, it should constitute the main target of acute stroke therapy. In a series of PET studies performed 5-18 h after stroke onset, we were able to (i) document, for the first time in man, the existence of tissue fulfilling operational criteria for penumbra in about one third of the cases; (ii) show that long-term neurological recovery is proportional to the volume of penumbra that eventually escapes infarction, and (iii) detect penumbral tissue as late as 16 h after symptom onset in occasional patients, suggesting the therapeutic window may be protracted in such cases. Mapping the penumbra in the individual patient with neuroimaging procedures should allow to formulate a pathophysiological diagnosis, and thus to design a rational management of the stroke patient and to improve the selection of candidates for therapeutic trials.