Ultrasound perfusion imaging (UPI) reliably detects size and localization of acute stroke. It remains unclear which time window detects, most sensitively and specifically, early changes of cerebral perfusion patterns and whether region-wise analysis is superior to parametric imaging analysis. Bilateral phase inversion harmonic imaging examinations (bolus kinetic, fitted model function) were performed twice (acutely and 28 h later) in 10 patients with acute ischemic stroke (<12 h). Examinations were evaluated using a region-wise analysis of the time-intensity curve and by parametric images of the time-to-peak intensity maps. Results were correlated in-between the ultrasound examinations and to follow-up cranial computed tomography (CCT) scans. Correlation between the early region-wise UPI examination and follow-up CCT was the strongest (Spearman correlation coefficient 0.76, sensitivity 84%, specificity 96%). Spearman coefficient between the late UPI examination and CCT was 0.51; sensitivity and specificity were 71% and 82%. Values in between UPI examinations were 57% and 88%, with a Spearman coefficient of 0.47 (p for all < 0.001). Values of the analysis of the parametric images were less strong. Concordance between both of the UPI methods was 65% in the early examination and 72% in the late examination. Changes of perfusion patterns are most accurately detected in the early course of stroke, when core of infarction can be differentiated from penumbra and viable tissue. Reperfusion phenomena may impair the diagnostic impact in later examinations. Parametric imaging does not yet seem to be as accurate as region-wise analysis.