Currently, a reliable method for real-time prediction of ischemia in the human brain is not available. Here, we took a first step towards validating non-invasive intraoperative laser speckle imaging (iLSI) for prediction of infarction in 22 patients undergoing decompressive surgery for treatment of malignant hemispheric stroke. During surgery, cortical perfusion was visualized and recorded in real-time with iLSI. The true morphological infarct extension within the iLSI imaging field was superimposed onto the iLSI blood flow maps according to a postoperative MRI (16 h [95% CI: 13, 19] after surgery) with three-dimensional magnetization-prepared rapid gradient-echo and diffusion-weighted imaging reconstruction. Based on the frequency distribution of iLSI perfusion values within the infarcted and non-infarcted territories, probability curves and perfusion thresholds of normalized cerebral blood flow predictive of eventual infarction or non-infarction were calculated. Intraoperative LSI predicted and excluded cortical ischemia with 95% probability at normalized perfusion levels below 40% and above 110%, respectively, which represented 73% of the entire cortical surface area. Together, our results suggest that iLSI is valid for (pseudo-) quantitative assessment of blood flow in the human brain and may be used to identify tissue at risk for infarction at a given time-point in the course of ischemic stroke.