We have used magnetic resonance imaging (MRI) techniques to characterise a rat model of thromboembolic stroke. The consequences of acute perfusion deficit associated with a middle cerebral artery occlusion (MCAo) by a newly formed thrombus was mapped by interrogation of the tissue oxygenation status using gradient echo methods and production of T2* maps. Final infarct size was subsequently assessed at 24-h post-ischaemia by histology with 2,3,5-triphenyltetrazolium chloride (TTC) staining. Animals displayed an infarct volume of 178.7+/-84.2 mm 3 (mean+/-S.D.) with a large coefficient of variation (47%) and range of values (85.6-265.5 mm 3 ). This variability provided us with an opportunity to assess the relationships between early imaging observations and eventual infarct size. For a single cerebral slice, at the centre of the MCA territory, a relationship between the area of reduced T2* at 1 and 2 h post MCAo correlated highly with final lesion area (Spearman rank correlation, r=0.98, P<0.01, n=9). Lesion volumes in the thromboembolic MCAo model were compared with a 120-min occlusion, 22-h reperfusion protocol using an intraluminal thread MCAo approach. For the thromboembolic model, the total lesion volume was found to be smaller (178.7+/-84.2 vs. 243.3+/-50.1 mm 3 , mean+/-S.D., Student's t-test P=0.046) and showed a greater variability (coefficient of variations: 47% vs. 21%). These data underline the relative variability of this embolic model and provide important preliminary information regarding the value of early changes in T2* in predicting eventual infarct size.