Objective: We aimed to establish the cut-off point of the Montreal Cognitive Assessment (MoCA-Beijing) in screening for cognitive impairment (CI) within 2 weeks of mild stroke or transient ischaemic attack (TIA). Methods: A total of 80 acute mild ischaemic stroke patients and 22 TIA patients were recruited. They received the MoCA-Beijing and a formal neuropsychological test battery. CI was defined by 1.5 SD below the established norms on a formal neuropsychological test battery. Results: Most stroke and TIA patients were in their 50s (53.95 +/- 11.43 years old), with greater than primary school level of education. The optimal cut-off point for MoCA-Beijing in discriminating patients with CI from those with no cognitive impairment (NCI) was 22/23 (sensitivity 85%, specificity 88%, positive predictive value=91%, negative predictive value=80%, classification accuracy=86%). The predominant cognitive deficits were characteristic of frontal-subcortical impairment, such as visuomotor speed (46.08%), attention/executive function (42.16%) and visuospatial ability (40.20%). Conclusions: A MoCA-Beijing cut-off score of 22/23 is optimally sensitive and specific for detecting CI after mild stroke, and TIA in the acute stroke phase, and is recommended for routine clinical practice.