Ischemic penumbra was first defined by Astrup in 1981 as perfused brain tissue at a level within the thresholds of functional impairment and morphological integrity, which has the capacity to recover if perfusion is improved. It exists, even for a short period of time in the center of ischemia, from which irreversible necrosis propagates to the neighboring tissues over time. Penumbra has become the focus of intense imaging research to differentiate it from infarction. Accurate detection of this 'tissue at risk' could be used to identify patients who would benefit most from acute treatment. Currently, recombinant tissue plasminogen activator (rtPA) is the only approved drug that has shown significant benefits in acute stroke patients when administered intravenously less than 4.5 h after stroke. However, its use is limited. Discrimination between infarct core and the surrounding potentially salvageable tissue is useful to better identify patients suitable for treatment. This can be achieved by positron emission tomography, single-photon-emission computed tomography, computed tomography perfusion scan and perfusion-weighted and diffusion-weighted magnetic resonance imaging. Identification of the penumbra might enable selective rtPA use in patients with large penumbras and small infarct cores, even beyond the 4.5-hour time window, where the penumbra may persist for more than 12 h. The purpose of this review was to describe neuroimaging modalities capable of identifying penumbra tissue so as to provide surrogate markers for new trials in acute ischemic stroke patients.