Background and Purpose: Neurological deterioration following acute stroke is common and associated with increased morbidity and mortality. The underlying pathophysiological mechanisms are not fully understood, and it is difficult to predict which patients are at risk of deterioration. Our study aimed to assess if acute MRI findings could be used for the prediction of stroke in progression (SIP). Methods: Prospectively 41 patients, 13 with lacunar infarcts and 28 with territorial infarcts, were admitted to an acute stroke unit within 24 h of stroke onset (median 11 h, range 3– 22). Diffusion-weighted imaging (DWI), perfusion-weighted imaging and magnetic resonance angiography were performed 3 times, immediately after clinical evaluation, on day 7 and after 3 months. Clinical neurological assessments were performed every 2 h during the first 24 h and once daily from day 2 to 7. SIP was defined as a permanent decrease of ≧3 Scandinavian Stroke Scale (SSS) points for speech or ≧2 SSS points for consciousness or ≧2 SSS points for limb strength, when assessed at baseline compared to the day after admission and daily during the following week. Patients were followed up on day 90 and assessed using the modified Rankin Scale, Barthel Index and SSS score. Patients with and without SIP were compared using both clinical and MRI data obtained on admission, on day 7 and after 3 months. Results: Fifteen patients (37%) developed SIP. Increased DWI lesion volume on day 7 in all strokes was associated with SIP (χ<sup>2</sup>, p = 0.005). All lacunar infarcts with a DWI volume >1.5 cm<sup>3</sup> at baseline (4 patients) developed SIP (p < 0.005). Patients with territorial infarcts and SIP had lower baseline SSS scores with severer symptoms than non-SIP patients (p ≤ 0.05). Despite trends in MRI differences at baseline, only the baseline SSS score in patients with territorial infarcts remained an independent predictor of SIP, when using a logistic regression model (p < 0.05). Conclusions: These findings suggest that SIP is due to growth of the lesion volume from baseline to day 7. In patients with lacunar infarcts, there was an association between initial lesion size and risk of developing SIP. However, for territorial infarcts, baseline SSS was found to be the best individual predictor of SIP and clinical outcome.