Multiparametric MRI including diffusion and perfusion imaging provides information on the extent of irreversibly damaged ischemic and/or critically hypoperfused tissue. Magnetic resonance angiography provides additional information on vessel status. The concept of perfusion-diffusion mismatch allows the estimation of tissue at risk of infarction which might be salvaged by timely reperfusion. In large case series and nonrandomized cohort studies, perfusion-diffusion mismatch-based thrombolysis was performed not less than 3 h after symptom onset with excellent safety and signs of good efficacy. However no randomised controlled trial has confirmed this to date. Recent studies improved the understanding of the mismatch concept and identified reperfusion unequivocally as an important predictor of the clinical response to thrombolysis. At the moment MRI-based thrombolysis can be performed after 3 h based on individual benefit:risk assessment in experienced stroke centers.