The selective cooling of severely injured brain tissue while maintaining normal temperature throughout the remaining body, to avoid cooling-related systemic side effects, has been proposed as a desirable method to improve the outcome of patients with acute brain catastrophes. One approach for targeted brain cooling may utilize miniature cooling probes directly inserted into injured brain tissue. Based on experimental data obtained in primates with normal and injured brains, this study simulates the expected temperature distributions surrounding a prototype brain cooling probe. Our model employs the Pennes bioheat equation to define the effects of local brain perfusion rate on the temperature field within brain tissue. Cooling penetration achieved by this probe under normal and globally ischemic conditions extended from 10 mm to 25 mm, respectively, from the device surface into the surrounding brain parenchyma, and was strongly dependent on the local brain perfusion, with a larger cooling penetration being obtained in injured (less perfused) brain regions. Further, the simulated results indicate that transient brain temperature behavior is affected by both the initial perfusion rate and the blood perfusion response to tissue cooling. Assuming a constant local blood perfusion rate during cooling, our model predicts an established steady state temperature field within 16 min, though additional time may be needed if the blood perfusion rate keeps changing during the cooling. It is also concluded that the brain cooling rate monitored by a temperature sensor close to the device may not be the most accurate measure of cooling penetration, as this estimate neglects to consider key variables such as local blood perfusion rate, monitoring location, and time duration over which the cooling rate is calculated.