Background. Stroke can be a challenging diagnosis in an emergency-setting. We sought to determine whether copeptin may be a useful biomarker to differentiate between ischemic stroke (IS), transient ischemic attack (TIA), and stroke-mimics. Methods. In patients with suspected stroke arriving within 4.5 hours of symptom-onset, copeptin-levels were measured in initial blood-samples. The final diagnosis was adjudicated by vascular neurologists blinded to copeptin-values. Results. Of all 36 patients with available copeptin-values (median age 71 years, IQR: 54-76; 44% female), 20 patients (56%) were diagnosed with IS, no patient was diagnosed with hemorrhagic stroke, nine patients (25%) were diagnosed with TIA, and seven patients (19%) were stroke-mimics. Copeptin-levels (in pmol/L) tended to be higher in patients with IS [19.1 (11.2-48.5)] compared to TIA [9.4 (5.4-13.8)]. In stroke-mimics the range of values was extremely broad [33.3 (7.57-255.7)]. The diagnostic accuracy of copeptin for IS was 63% with a sensitivity of 80% and a positive predictive value of 64%. Conclusion. In this cohort of patients copeptin-levels within 4.5 hours of symptom onset were higher in patients with IS compared to TIA but the broad range of values in stroke-mimics limits diagnostic accuracy. This trial is registered with UTN: U1111-1119-7602.