Phenobarbital (PB) has been used at several pediatric centers for prophylaxis against neonatal hyperbilirubinemia. However, few attempts have been made to evaluate this procedure quantitatively, and a variety of dose schedules has been proposed. Therefore, a randomized, controlled clinical trial was performed in which the effects on bilirubin disposition and on neonatal behavior was quantitated. Forty‐three preterm infants were randomized into one of four dose groups and given 0, 4, 8, or 12 mg of PB per kg in a single dose within the first few hours after birth (mean 2.2 h). The total serum bilirubin disappearance rate was found to be significantly increased (p < 0.01) only in the 12 mg/kg group. This effect was not evident until postnatal day 7. The 4 and 8 mg/kg groups were not significantly different from the control group at any time. Infant behavior was monitored by a non‐invasive time‐lapse filming technique. The time spent in quiet sleep was found to be proportional to the plasma PB concentration at one day of age (r= 0.61). The infants in the 12 mg group spent a larger proportion of time in quiet sleep than the other groups (p < 0.05). The plasma half‐lives, plasma clearances and volumes of distribution of PB were similar in the three dose groups. No correlation was found between the pharmacokinetics and the gestational age of the infant. It is concluded that in order to enhance the bilirubin disappearance rate, PB has to be administered in doses that will affect behavior.