Objective: Intravenous immune globulin (IVIG) reduces jaundice in many but not all cases of neonatal isoimmunization. We sought to elucidate the type of infant most likely to benefit from IVIG administration by attempting to define pretreatment parameters associated with both clinical symptomatology and therapeutic responsiveness to IVIG. Methods: Term, healthy Coombs-positive infants were studied prospectively. IVIG was administered if, despite phototherapy, serum bilirubin reached ≥ 222 μmol/l (13 mg/dl) at ≤ 24 h of age and/or ≥ 274 μmol/l (16 mg/dl) at > 24 h of age. Clinical data including serial serum total bilirubin levels, rate of bilirubin rise on day 1 of life, serial corrected carboxyhemoglobin levels (a sensitive indicator of hemolysis) and total hemoglobin (tHb) levels were collected. Results: Infants were classified as IVIG responders (n = 18), those in whom total serum bilirubin levels either remained stable or decreased following IVIG administration; IVIG nonresponders (n = 5), those who developed a total serum bilirubin of ≥ 2 mg/dl greater than pre-IVIG bilirubin levels within the first 24 h after IVIG administration, or nontreated, those not meeting IVIG treatment criteria (n = 13). Four of the five nonresponders proceeded to require exchange transfusion vs. none of the others (p < 0.001). Four of the five nonresponders had a pretreatment rate of bilirubin rise of > 1 mg/dl/h as compared with only 1 of 18 responders and none of the nontreated (p < 0.001). Pretreatment tHb levels were also different (13.2 ± 1.3 vs. 15.5 ± 2.3 vs. 17.7 ± 2.4 g/dl for nonresponders vs. responders vs. nontreated infants, respectively; p < 0.005). The highest pretreatment COHbc levels were seen in the nonresponders (1.8 ± 0.7 vs. 1.4 ± 0.3 vs. 0.9 ± 0.3% tHb, respectively). Conclusions: Our 3 groups represent a spectrum of hemolysis, ranging from severe to moderate to mild. This spectrum appears to relate not only to the severity of hemolysis, but also to the therapeutic responsiveness to IVIG. We speculate that some or all of the factors identified can be used prospectively to predict the subsequent clinical course of ABO-incompatible infants and to facilitate optimal management.