The Canadian Neurological Scale (CNS) was designed to monitor mentation and motor functions in stroke patients. We assessed its validity and reliability on a group of 157 patients with a diagnosis of acute cerebrovascular accident. We determined validity by (1) correlating scale items and total score with the standard neurologic examination; (2) exploring the scale's predictive power with different end points at 6 months--the initial CNS was a significant predictor of outcome; (3) showing that the CNS had higher correlation coefficients with the initial neurologic examination than the Glasgow Coma Scale; and (4) assessing the responsiveness of the scale to change in the neurologic status of stroke patients. Interobserver reliability, measured by kappa statistics on each scale item, was good. Accordingly, we established the validity and reliability of the CNS for its use in clinical studies and in the care of stroke patients.