Summary Purpose The present study was undertaken to determine whether flushing the carotid artery with normal saline at 4 °C (hypothermic carotid arterial flush, HCAF) during cardiac arrest can achieve selective cerebral hypothermia rapidly during cardiac arrest and improve cerebral outcome. Methods Ventricular fibrillation (VF) was induced in fourteen dogs and circulatory arrest was maintained for 9 min. Dogs were then resuscitated by cardiopulmonary resuscitation. The dogs were divided into two groups; a control group ( n = 7), which underwent precisely the same procedure as the experimental group but not HCAF, and an experimental group (HCAF group; n = 7), which received HCAF from 8 min after the onset of VF. Results Two dogs in the control group and in the HCAF group died within 72 h after the recovery of spontaneous circulation (ROSC) due to extracerebral complications. The remaining 10 dogs survived to final evaluation at 72 h post-ROSC. In the HCAF group, tympanic temperature decreased from 37.7 °C (37.5–37.8) to 34 °C in 1 min (1–1.5) from the start of HCAF and was maintained below 34 °C until 6.5 min (3–12) after the start of HCAF, whereas oesophageal and rectal temperatures were maintained above 35 °C. Neurological deficit scores (0–100%) at 72 h post-ROSC were 42.4% (27.0–80.6) in the control group and 18.4% (14.0–36.0) in the HCAF group ( p < 0.05). Conclusion HCAF induced selective cerebral hypothermia rapidly during cardiac arrest and improved neurological deficit scores after 9 min of no blood flow in the described canine cardiac arrest model.