The prevalence of vitamin K deficiency in newborn infants and the influence of perinatal risk factors were studied prospectively in 934 infants. A noncarboxylated prothrombin assay to detect proteins induced in vitamin K absence (PIVKA-II) was used to determine the presence of vitamin K deficiency; of 934 cord blood samples assayed, 2.9% were positive for PIVKA-II (0.015 to 0.15 U/ml). All infants found to have detectable PIVKA-II were born at term. The number of infants positive for PIVKA-II was greater in the group small for gestational age (7.4%) than in those appropriate (2.7%) or large (3.1%) for gestational age. Nine categories of perinatal risk groups were defined: however, the majority of infants who were PIVKA-II positive (63%) were normal. All infants received prophylactic vitamin K, and no infant with PIVKA-II in the cord sample subsequently had clinical bleeding. In two patients the rate of 50% disappearance of PIVKA-II after vitamin K administration approximated 70 hours. Two PIVKA-II positive patients with active bleeding or disseminated intravascular coagulation had an accelerated disappearance of 20 to 40 hours. The long disappearance time of PIVKA-II in a steady state may allow detection of vitamin K deficiency despite administration of vitamin K. The majority of cases of neonatal vitamin K deficiency occurred in normal newborn infants. Therefore, all infants should receive prophylactic vitamin K at birth.