A prospective study was undertaken to evaluate the use of glycohemoglobin (HbA1) as a screening tool for the detection of gestational diabetes in 90 subjects. At 28 to 30 weeks, these patients had an HbA, determination prior to diabetes screening with a 1-hour serum glucose after 50 gm of oral glucose. Those women with an abnormal screening result were then evaluated with a standard 3-hour oral glucose tolerance test. Forty-seven women had an abnormal 1-hour serum glucose (greater than or equal to 140 mg/dl). Nine of these 47 women (19%) had an abnormal HbA1. Four of these nine women (44.4%) with an abnormal screening glucose as well as an abnormal HbA1 demonstrated an abnormal oral glucose tolerance test. Eleven of 38 women (29%) with an abnormal glucose screen but normal HbA1 had an abnormal glucose tolerance test. The combination of two abnormal tests, glucose screen and HbA1, predicted gestational diabetes in only four of 15 (27%) of our patients. The sensitivity and specificity of this combination were 22.4% and 87.8%, respectively. The median values of HbA, with normal and abnormal glucose tolerance tests were 7.8% (range, 4.9% to 11.5%) and 8.2% (range, 5.4% to 12%), respectively. These values were not significantly different. These data confirm that HbA, alone or in combination with a 50 gm glucose screen is not a sensitive screening tool for the detection of gestational diabetes mellitus.