Hypothermia after perinatal hypoxia-ischemia (HI) is neuroprotective; the precise brain temperature that provides optimal protection is unknown. To assess the pattern of brain injury with 3 different rectal temperatures, we randomized 42 newborn piglets: (Group i) sham-normothermia (38.5-39 degrees C); (Group ii) sham-33 degrees C; (Group iii) HI-normothermia; (Group iv) HI-35 degrees C; and (Group v) HI-33 degrees C. Groups iii through v were subjected to transient HI insult. Groups ii, iv, and v were cooled to their target rectal temperatures between 2 and 26 hours after resuscitation. Experiments were terminated at 48 hours. Compared with normothermia, hypothermia at 35 degrees C led to 25 and 39% increases in neuronal viability in cortical gray matter (GM) and deep GM, respectively (both p < 0.05); hypothermia at 33 degrees C resulted in a 55% increase in neuronal viability in cortical GM (p < 0.01) but no significant increase in neuronal viability in deep GM. Comparing hypothermia at 35 and 33 degrees C, 35 degrees C resulted in more viable neurons in deep GM, whereas 33 degrees C resulted in more viable neurons in cortical GM (both p < 0.05). These results suggest that optimal neuroprotection by delayed hypothermia may occur at different temperatures in the cortical and deep GM. To obtain maximum benefit, you may need to design patient-specific hypothermia protocols by combining systemic and selective cooling.