In this study, we aimed to assess the detection of crossed cerebellar diaschisis (CCD) following stroke by perfusion-weighted magnetic resonance imaging (PW-MRI) in comparison with positron emission tomography (PET). Both PW-MRI and 15O-water-PET were performed in acute and subacute hemispheric stroke patients. The degree of CCD was defined by regions of interest placed in the cerebellar hemispheres ipsilateral (I) and contralateral (C) to the supratentorial lesion. An asymmetry index (AI = C/I) was calculated for PET-cerebral blood flow (CBF) and MRI-based maps of CBF, cerebral blood volume (CBV), mean transit time (MTT), and time to peak (TTP). The resulting AI values were compared by Bland-Altman (BA) plots and receiver operating characteristic analysis to detect the degree and presence of CCD. A total of 26 imaging procedures were performed (median age 57 years, 20/26 imaged within 48 hours after stroke). In BA plots, all four PW-MRI maps could not reliably reflect the degree of CCD. In receiver operating characteristic analysis for detection of CCD, PW-CBF performed poorly (accuracy 0.61), whereas CBV, MTT, and TTP failed (accuracy < 0.60). On the basis of our findings, PW-MRI at 1.5 T is not suited to depict CCD after stroke.