Background There is still a considerable delay between the onset of symptoms and arrival at a stroke unit for most patients with acute stroke. The aim of the study was to describe the feasibility of a pre-hospital diagnosis of stroke by an emergency medical service (EMS) nurse in terms of diagnostic accuracy and delay from dialing 112 until arrival at a stroke unit. Methods Between September 2008 and November 2009, a subset of patients with presumed acute stroke in the pre-hospital setting were admitted by EMS staff directly to a stroke unit, bypassing the emergency department. A control group, matched for a number of background variables, was created. Results In all, there were 53 patients in the direct admission group, and 49 patients in the control group. The median delay from calling for an ambulance until arrival at a stroke unit was 54 minutes in the direct admission group and 289 minutes in the control group (p < 0.0001). In a comparison between the direct admission group and the control group, a final diagnosis of stroke, transient ischemic attack (TIA) or the sequelae of prior stroke was found in 85% versus 90% (NS). Among stroke patients who lived at home prior to the event, the percentage of patients that were living at home after 3 months was 71% and 62% respectively (NS). Conclusions In a pilot study, the concept of a pre-hospital diagnosis of stroke by an EMS nurse was associated with relatively high diagnostic accuracy in terms of stroke-related diagnoses and a short delay to arrival at a stroke unit. These data need to be confirmed in larger studies, with a concomitant evaluation of the clinical consequences and, if possible, the level of patient satisfaction as well.