According to the Pedersen hypothesis, fetal hyperinsulinism is the major cause for adverse neonatal outcome. We investigated associations between insulin levels in cord blood and fetal complications. Three groups of 21 insulin-dependent diabetic patients with different insulin levels in cord blood were matched according to White Classes. Insulin levels in cord blood of < 20 microU/ml were considered normal (controls), 20-50 microU/ml intermediate group, and > 50 microU/ml high (cases). The mean (+/-S.D.) insulin level in cord blood in the three groups was 10.7+/-5.6, 28.6+/-8.1, and 104.0+/-61.0 microU/ml, respectively. Controls and cases showed significant differences in birth weight > 90th percentile (9.5% vs. 76.2%), premature birth < 37 weeks (4.8% vs. 71.4%), caesarean delivery (28.6% vs. 66.4%), hypoglycaemia of the neonate (14.3% vs. 61.9%), cushingoid appearance (4.8% vs. 42.9%) and respiratory distress syndrome (0% vs. 33.3%). The results of the intermediate group were between the controls and the cases. Insulin levels in cord blood > 20 microU/ml represent a continuum of increasing diabetogenic fetopathy. We consider neonates with insulin levels in cord blood < 20 microU/ml as metabolically healthy, those with 20-50 microU/ml as having mild fetopathy, and those with > 50 microU/ml as having marked fetopathy, respectively.