Background: To systematically evaluate retinal diffusion restrictions (RDR) in patients with central retinal artery occlusion (CRAO) using a standard stroke DWI-EPI sequence. Methods: CRAO patients admitted between 01/2010 to 12/2019 and MRI performed within 2 weeks of clinical onset were included in this retrospective cohort study. Evaluation of diffusion-weighted imaging for retinal diffusion restrictions was performed by a neuroradiologist blinded for CRAO side and clinical data. The influence of selected clinical and technical MRI parameters on the presence of RDR in CRAO was assessed using Chi-squared statistics and Bonferroni post-hoc analysis. Results: 127 patients (69,6 ± 13,9 years; 59 female) with 131 DWI scans were included in the study. Overall sensitivity of RDR in CRAO was 67,2%. RDR were falsely attributed to the wrong eye in a single case only (0,8%). RDR in CRAO were reliably identified up to 1 week after onset of amaurosis with highest sensitivity in DWI performed within 24 hours (79%). Contrariwise, detection rates of RDR dropped significantly in the second week (10,0%; p=0,0006). Absence of RDR was more prevalent in patients without fundoscopic presence of retinal edema (60% vs. 27,1%; p = 0,004) and in subjects with restitution of visual acuity at discharge (75% vs. 28,4%; p = 0,006). Conclusions: RDR in CRAO can reliably be identified in a majority of CRAO patients with standard stroke DWI performed within 24 hours to 1 week after onset of amaurosis. Further studies are needed to investigate the potential of stroke DWI in hyperacute CRAO and the utility of RDR in selecting patients for thrombolytic therapy.