The urgent need to recognize infants which are at high risk for developing significant jaundice is apparent in the era of routine early discharge. The aim of the present study was to determine which factors are independently associated in the healthy term newborn with neonatal jaundice (defined as total serum bilirubin (TSB) > 10.0 mg/dl at day 2, 214.0 mg/dl at day 3, > 17.0 mg/dl at days 4 and 5 of life). We prospectively followed 1,117 healthy term newborns born in two university affiliated community hospitals in Jerusalem, from October 29 through December 31, 1995. Using a multiple logisitic regression analysis, neonatal jaundice was best predicted(p<0.0001) by day 1 TSB. The adjusted odds ratio was 3.1 for each increase of 1 mg/dL TSB (95% confidence limits 1.4 - 4.1), and change in TSB from the first to the second day of life [2.4 (per mg/dl) (1.9 - 3.0)]. Maternal blood type O [2.9 (1.5 - 5.8)] age [1.1 (per year) (1.0 - 1.2)], schooling [0.8 (per year) (0.7 - 0.9)], and full breast-feeding [0.4 (0.2 - 0.9)] were also associated with jaundice (p<0.005). Other factors considered in the regression model, but found not significantly related to jaundice, induced maternal ethnic origin, smoking, hypertension, diabetes mellitus, intranatal administration of oxytocin, meperidine, anesthesia, premature rupture of the membranes, parity, newborn sex, birth weight, postnatal weight loss, gestational age, presentation, Apgar scores, blood type, hematocrit, Coombs' test, Glucose-6-phosphate deficiency, cephalohematoma, and history of jaundice in other siblings. A model for predicting neonatal jaundice based on the above factors has a sensitivity of 81.8%, a specificity of 82.9%, a false positive rate of 80.2% and a false negative rate of 1.1%. We conclude that individual risk assessment upon discharge is of value in directing more intensive follow-up toward infants at greater risk for neonatal jaundice.