OBJECTIVE: Prior studies have suggested that macrosomia is the only morbid condition associated with gestational diabetes and that this association is the result of confounding by maternal obesity rather than a result of gestational diabetes itself. We sought to determine whether unrecognized gestational diabetes is an independent predictor of macrosomia and other perinatal morbid conditions after controlling for confounding variables. STUDY DESIGN: A retrospective analysis of 472 consecutive cases of gestational diabetes diagnosed between 24 and 30 weeks' gestation was undertaken including 16 prospectively identified but clinically unrecognized cases, 297 cases treated with diet alone, and 76 treated with diet plus insulin. Unrecognized cases were matched to 64 nondiabetic controls for race, age, body mass index, parity, pregnancy weight gain, and gestational age at delivery. RESULTS: In the unrecognized gestational diabetes group versus the nondiabetic control versus gestational diabetes diet groups rates of large for gestational age infants (44% vs 5% vs 9%, p < 0.0005), macrosomia (44% vs 8% vs 15%, p < 0.01), shoulder dystocia (19% vs 3% vs 3%, p < 0.05), and birth trauma (25% vs 0% vs 0.3%, p < 0.001) were all significantly increased. These differences remained significant after controlling for maternal age, race, parity, body mass index, pregnancy weight gain, and gestational age at delivery. CONCLUSIONS: This study suggests that unrecognized gestational diabetes increases risks of large for gestational age infants, macrosomia, shoulder dystocia, and birth trauma independent of maternal obesity and other confounding variables. Clinical recognition and dietary control of gestational diabetes are associated with a reduction in these perinatal morbid conditions. (Am J Obstet Gynecol 1998;178:1321-32.)