OBJECTIVE: To determine how well parents, nurses, physicians, and an Ingram icterometer can detect the presence and the severity of jaundice in newborns. SETTING: Normal newborn nursery in a 340-bed teaching hospital. PATIENTS OR OTHER PARTICIPANTS: Nurses and physicians caring for nursery infants and parents of the infants. INTERVENTIONS: Physicians and nurses examining newborns documented whether they detected jaundice in the infants and, if so, the estimated bilirubin level and the extent of cephalocaudal progression of the jaundice. An assistant taught the parents how to examine the infants for jaundice and determine its cephalocaudal progression. The assistant also obtained icterometer readings. Bilirubin testing was performed according to usual clinical practice. OUTCOME MEASURES: Nurse and physician estimates of bilirubin levels; parent, nurse, and physician assessment of the presence of jaundice and its cephalocaudal progression; icterometer readings; bilirubin levels. RESULTS: There was moderate agreement about the presence of jaundice in the infants (pairwise kappa, 0.48) However, all infants with bilirubin levels >12 mg/dL were correctly identified as jaundiced by all examiners. The parents' assessment of cephalocaudal progression and the icterometer readings were most highly correlated with serum bilirubin levels (adjusted Pearson correlations, 0.71 and 0.57, respectively). CONCLUSIONS: Many parents can be taught to accurately assess cephalocaudal progression of jaundice in the hospital. The icterometer is a useful tool for assessing jaundice severity. Both parent assessment and the icterometer were more highly correlated with bilirubin levels than physician and nurse estimates in this study. Additional research is needed to determine how accurate these methods of clinical assessment are at the higher bilirubin levels that typically occur after hospital discharge.