Aims. To determine the practicality and safety of head cooling with mild or minimal systemic hypothermia in term neonates with moderate to severe hypoxic-ischemic encephalopathy. Methods. Study group infants ≥37 weeks' gestation, who had an umbilical artery pH ≤7.09 or Apgars ≤6 at 5 minutes, plus evidence of encephalopathy. Infants with major congenital abnormalities were excluded. Trial Design. Infants were randomized to either no cooling (controls; rectal temperature = 37.0 ± 0.2°C,n = 10) or sequentially, either minimal systemic cooling (rectal temperature = 36.3 ± 0.2°C,n = 6) or mild systemic cooling (rectal temperature = 35.7 ± 0.2°C, n = 6). Head cooling was accomplished by circulating water at 10°C through a coil of tubing wrapped around the head for up to 72 hours. All infants were warmed by servo-controlled overhead heaters to maintain the allocated rectal temperature. The rectal, fontanelle, and nasopharyngeal temperatures were continuously monitored. Results. From January 1996 to October 1997, 22 term infants were randomized from 2 to 5 hours after birth. All infants showed a metabolic acidosis at delivery, with similar umbilical artery pH in the control group (mean ± standard deviation, 6.79 ± 0.25), minimal cooling group (6.98 ± 0.21), and mild cooling group (6.93 ± 0.11), and depressed Apgar scores at 5 minutes in the control group (4.5 ± 2), minimal cooling group, (4.7 ± 2) and mild cooling group (6.0 ± 1). In the mild-cooled infants, the nasopharyngeal temperature was 34.5°C during cooling, 1.2°C lower than the rectal temperature. This gradient narrowed to 0.5°C after cooling was stopped. No adverse effects because of cooling were observed. No infants developed cardiac arrhythmias, hypotension, or bradycardia during cooling. Thrombocytopenia occurred in 2 out of 10 controls, 2 out of 6 minimal cooling infants, and 1 out of 6 mild cooling infants. Hypoglycemia (glucose <2.6 mM) was seen on at least one occasion in 2 out of 10 controls, 4 out of 6 minimal cooling infants, and 1 out of 6 mild cooling infants. Acute renal failure occurred in all infants. The metabolic acidosis present in all infants at the time of enrollment into the study progressively resolved despite cooling, even in the mild hypothermia group. Conclusions. Mild selective head cooling combined with mild systemic hypothermia in term newborn infants after perinatal asphyxia is a safe and convenient method of quickly reducing cerebral temperature with an increased gradient between the surface of the scalp and core temperature. The safety of mild hypothermia with selective head cooling is in contrast with the historical evidence of adverse effects with greater depths of whole-body hypothermia. This safety study and the strong experimental evidence for improved cerebral outcome justify a multicenter trial of selective head cooling for neonatal encephalopathy in term infants.