INTRODUCTION American Stroke guidelines suggest that oxygen saturation after acute stroke should be maintained at 95% or higher. European recommendations include a dose of 2 to 4 L/minute via nasal cannulae. These recommendations are not based on controlled clinical studies. No consensus exists among stroke physicians regarding how much oxygen should be administered and which route should be selected. The aim of this study was to assess the effects of different doses and routes of oxygen administration on oxygen saturation in patients with stroke. METHODS Patients were recruited within 72 hours of admission with acute stroke. Oxygen saturation was assessed by pulse oximetry (Minolta Pulsox 3i) for 30 minutes on room air (baseline), 2 and 3 L/minute of oxygen via nasal cannulae, 24 and 35% oxygen via face mask, and, finally, on room air. Patients were asked about their preferred route of oxygen administration. RESULTS Twenty-one patients with a mean age of 72.3 years (range: 46-87 years) were recruited. A total of 24% oxygen (face mask) increased saturation by 1%, 2 L/minute (nasal cannulae) and 35% (face mask) led to a 2% increase, and 3 L/minute (nasal cannulae) resulted in a 3% increase in saturation compared to baseline (p < 0.001). Oxygen saturations were similar in the first and last half-hours. Oxygen treatment has no significant effect on respiratory and heart rates. Most patients (71%) preferred nasal cannulae to face masks. CONCLUSIONS In our patient population, there was a close dose-response relationship between the amount of oxygen given and the resultant changes in oxygen saturation.