The current practice for treating neonates for jaundice centers on the recommendation that bilirubin levels should be kept below 20 mg/dL. Preventing bilirubin levels from exceeding 20 mg/dL, however, does not guarantee the avoidance of kernicterus, lower IQs or neurologic abnormalities. Studies in the 1960s and 1970s reported cases of infants with clinical and pathological kernicterus whose neonatal bilirubin levels were well below 20 mg/dL. It is now well accepted that protein binding, acidosis, hypoxia, intracranial hemorrhage and hemolytic disease play a role in facilitating bilirubin toxicity. This paper reviews previously published studies that were instrumental in identifying the role of hypoxia, acidosis, hemolytic disease, intracranial hemorrhage and protein binding in bilirubin encephalopathy and identifies two key variables which contribute to bilirubin flux-free bilirubin concentration and time. The paper proposes a new approach for evaluating bilirubin levels termed 'bilirubin index'. Future research should initially focus on healthy term infants without concomitant illness and should record free bilirubin levels as a function of time. The area under the bilirubin versus time curve represents the integration of bilirubin level with respect to time, or simply termed the 'bilirubin index'. The bilirubin index could then be correlated with parameters for measuring neurological outcome. Assuming a correlation would exist, the bilirubin index may then become the number for guidance with respect to intervention therapy. Attempting to address this issue by starting with a healthy population of neonates and correlating bilirubin index with neurological outcome offers a better chance for uncovering that 'threshold of toxicity'.