OBJECTIVE In this study, we investigated the lacunar hypothesis to answer three questions: 1) Is the lacunar syndrome valid for diagnosing lacunar infarction? 2) What is the frequency of potential cardiac versus carotid sources of embolism in patients with lacunar versus cortical infarct? 3) What is the frequency of vascular risk factors in these two groups of patients? METHODS The study was performed in a well-defined prospective series of 103 patients with a first-ever lacunar infarct and 144 other patients with a first-ever infarct involving the cortex. RESULTS Sensitivity and specificity of the lacunar syndromes in diagnosing lacunar infarction were 95% and 93%, respectively. Positive and negative predictive values of diagnosing lacunar infarction in patients with lacunar syndromes were 90% and 97%, respectively. Risk factor analysis showed no differences for either group of cerebral infarction. A cardiac source of embolism was significantly less frequent in patients with lacunar infarction (odds ratio = 0.32, 95% confidence interval = 0.17-0.61, p less than 0.001). Significant carotid stenosis (diameter reduction greater than or equal to 50%) was also less frequent in patients with lacunar infarction (odds ratio = 0.35, 95% confidence interval = 0.16-0.76, p less than 0.001). CONCLUSIONS These findings show that the lacunar syndrome is an excellent clinical test for diagnosing lacunar infarction and that cardiac and carotid embolism are unlikely causes of lacunar infarction, supporting the hypothesis that lacunar infarcts are usually caused by small vessel disease.