Background: Methods to assess impaired consciousness in acute stroke typically include the Glasgow Coma Scale (GCS), but the verbal component has limitations in aphasic or intubated patients. The FOUR (Full Outline of UnResponsiveness) score, a new coma scale, evaluates 4 components: eye and motor responses, brainstem reflexes and respiration. We aimed to study the interobserver variability of the FOUR score in acute stroke patients. Methods: We prospectively enrolled consecutive patients with acute stroke admitted from February to July 2008 to the stroke unit of our Neurology Department. Patients were evaluated by neurology residents and nurses using the FOUR score and the GCS. For both scales, we obtained paired and total weighted kappa values (Kw) and intraclass correlation coefficients (ICC). NIH stroke scale was also recorded on admission. Results: We obtained a total of 75 paired evaluations in 60 patients (41 cerebral infarctions, 15 cerebral hemorrhages and 4 transient ischemic attacks). Thirty-three (55%) patients were alert, 17 (28.3%) drowsy and 10 (16.7%) stuporous or comatose. The overall rater agreement was excellent in the FOUR score (Kw 0.93; 95% CI 0.89–0.97) with an ICC of 0.94 (95% CI 0.91–0.96) and in the GCS (Kw 0.96; 95% CI 0.94–0.98) with an ICC of 0.96 (95% CI 0.93–0.97). A good correlation was found between the FOUR score and the GCS (ρ 0.83; p < 0.01) and between the FOUR score and the NIH stroke scale (ρ –0.78; p < 0.001). Conclusions: The FOUR score is a reliable scale for evaluating the level of consciousness in acute stroke patients, showing a good correlation with the GCS and the NIH stroke scale.