Stroke therapy trials have historically allowed for late patient entry (ie, within 24 to 48 hours from stroke onset) despite evidence suggesting the importance of early intervention. Experimental studies of cerebral infarction suggest treatment may be most effective when begun within three hours and may be only marginally effective when begun after 12 hours. Lysis of an acute intra-arterial thrombus in the setting of thrombolytic therapy is also time dependent. We describe an ongoing dose-escalation study of tissue plasminogen activator (tPA) as ultra-early therapy for cerebral infarction. The protocol requires that hemorrhage be ruled out by computed tomography scan of the brain prior to tPA infusion, and the infusion must begin within 90 minutes of symptom onset. The two primary goals of the study are to assess safety and potential efficacy. Preliminary results from the study and the future of ultra-early stroke intervention are discussed.