Background: Stroke is one of the leading causes of death and disability worldwide. The diagnosis of stroke remains largely clinical, yet widely used stroke scoring systems and brain imaging do not satisfactorily allow the distinction of ischaemic stroke (IS) patients from stroke mimics (SMs). Blood biomarkers are promising tools that could facilitate clinical triage. Methods: This study recruited 66 patients with IS and 24 SMs. The levels of Glial fibrillary acidic protein (GFAP), Neuron-specific enolase (NSE), Neurofilament light chain (NfL) and blood-brain barrier (BBB) proteins [Occludin (OCLN), Zonula occludens 1 (ZO-1), Claudin-5] in blood serum were measured by enzyme-linked immunosorbent assay technique. Biomarker levels in IS patients and SMs were compared using the Mann–Whitney U test. Multivariable logistic regression analysis was used to evaluate the diagnostic performance of biomarkers in combination with the National Institutes of Health Stroke Scale (NIHSS) score. Results: More significant differences in circulating GFAP, NfL, OCLN, ZO-1, and Claudin-5 but not NSE were found in IS patients compared to SMs. A combination of circulating ZO-1, Claudin-5, and OCLN with NIHSS score gives the highest diagnostic accuracy, sensitivity, and specificity. Conclusions: A prediction model with circulating BBB proteins in combination with NIHSS score differentiates between IS patients and SMs.