The American Academy of Pediatrics has proposed guidelines for treating term/near term infants with hyperbilirubinemia based primarily on maintaining the total serum bilirubin concentration (TSB) below a "critical" level of 25 mg/dL (426 micromol/L). We estimated the sensitivity and specificity of this critical TSB using patient data from published reports. A TSB >25 mg/dL is recorded in about 92% of term/near term infants with kernicterus. When TSB is adjusted lower for risk factors of prematurity, sepsis, and isoimmune hemolytic disease, the guidelines have nearly a 98% sensitivity. The specificity of the critical TSB, however, is very poor; false positive rates exceed 90% if all cases of acute bilirubin toxicity are included, and about 94-96% if the endpoint is mild or severe residual brain injury. Clinically measurable confounders that might explain the large variance in critical TSB include the plasma unbound "free" bilirubin concentration (Bf) and, possibly, the concentration of bilirubin photoisomers. Limited clinical experience supports the proposition that Bf is superior to TSB in identifying patients at risk. Specificity of Bf and TSB was compared in small cohort of babies with acute and permanent bilirubin encephalopathy. The false positive rate for TSB (at 100% sensitivity) was 94%, compared with only 55% for Bf. A more comprehensive comparison of TSB and Bf, controlled for clinical confounders and photoiomers, is needed to formulate intervention guidelines with improved specificity that will reduce hospital admissions for unnecessary treatment.