&NA;We have investigated the effectiveness of surface cord cooling in reducing spinal cord temperature with respect to the rate of cooling and the temperature gradient within the spinal cord parenchyma. Another question of some practical importance if spinal cooling is to be used clinically is that of the temperature within the spinal cord as a function of the dura being intact or open. In five monkeys, a laminectomy was carried out at T‐10 and an impact injury of 350 g‐cm force was applied to the spinal cord with the dura intact. Hypothermic perfusion with Elliott's B solution (artificial cerebrospinal fluid) was started, and we measured temperatures in the spinal cord with a thermistor probe mounted on a stereotactic drive. A series of measurements were made with the dura intact and the measurements were then repeated with the dura open. For comparison, we also recorded temperatures in one animal that had not been cord‐injured. The rate of cord cooling was rapid during the first 3 minutes of hypothermic perfusion, after which there was a slight further reduction in cord temperature; a low level plateau of 6.7° C was reached within 21 minutes after the start of cooling. The temperature gradient at varying depths of the spinal cord was approximately 6° C (3.8° C at the posterior surface to 10.1° C in the deepest portion of the spinal cord), representing a temperature gradient of 1.3° C/mm of cord tissue. Transmission of the cooling effect from perfusate to the spinal cord was not appreciably affected by the dura being intact or open. (Neurosurgery,5: 583‐587, 1979)