Rapid focal cooling is an attractive nondestructive strategy to control and possibly prevent focal seizures. However, the temperature threshold necessary to abort seizures in primates is still unknown. Here, we explored this issue in a primate epilepsy model and observed the effect of rapid cooling on different electroencephalogram frequency bands, aiming at providing necessary experimental data for future clinical translational studies and exploring the mechanism of focal cooling in terminating seizures. We induced focal neocortical seizures using microinjection of 4-aminopyridine into premotor cortex in five anesthetized cynomolgus monkeys. The rapid focal cooling was implemented by using a thermoelectric (Peltier) device. As a result, the average durations of seizures and interictal intervals before cooling were 94.3 ± 4.0 s and 62.3 ± 6.9 s, respectively. When the cortex was cooled to 20 °C or 18 °C, there was no effect on seizure duration (109.4 ± 30.0 s, 91.3 ± 19.3 s) or interictal duration (99.4 ± 26.8 s, 83.2 ± 11.5 s, P > 0.05). But when the cortex was cooled to 16 °C, the seizure duration was reduced to 54.1 ± 4.9 s and the interictal duration was extended to 175.0 ± 16.7 s (P < 0.05). Electroencephalogram spectral analysis showed that the power of delta, alpha, beta, gamma and ripples bands in seizures were significantly reduced at 20 °C and 18 °C. At 16 °C, the power of theta band in seizures was also significantly reduced along with the other bands. Our data reveal that the temperature threshold in rapid focal cooling required to significantly shorten neocortical seizures in nonhuman primates is 16 °C, and inhibition of electroencephalogram broadband spectrum power, especially power of theta band, may be the underlying mechanism to control seizures.