Advances in neuroimaging have been central to the expansion of knowledge in the neurosciences over the past 20 years. One of the most important roles of brain imaging is in the selection of patients for acute stroke therapy. Currently, computed tomography (CT) is commonly used to select patients who have had strokes for thrombolytic therapy on the basis of the absence of haemorrhage and, more controversially, the presence of early CT changes of ischaemia. Since patients with ischaemic penumbra are more likely than those without to respond to therapy, identification of patients with this feature will become increasingly important. Although several imaging modalities can identify the penumbra, the most practical is magnetic resonance imaging (MRI) showing perfusion-weighted and diffusion-weighted imaging mismatch. Although uncertainties in image interpretation remain, surrogate MRI outcome measures are becoming an important component of translational research. Future developments in imaging technologies may provide other opportunities for surrogate outcome studies.