The goal of this study was to determine whether serum glycosylated protein levels (i.e., fructosamine) can reliably screen for gestational diabetes and whether these levels are valid markers of short-term glycemic control in the third trimester of pregnancy. Ninety-seven pregnant women at 26-28 wk gestation were evaluated over 9 mo. HbA1c and serum glycosylated protein (serum fructosamine) were determined at the baseline venipuncture of the 100-g oral glucose tolerance test performed to detect gestational diabetes. Of the 97 women studied, 13 tested positive for gestational diabetes (National Diabetes Data Group criteria). There were significant differences in the fasting and 1-, 2-, and 3-h glucose values between nondiabetic and diabetic patients (P less than 0.005 at each time point). No difference was noted in the baseline serum glycosylated protein level (2.02 +/- 0.08 vs. 1.98 +/- 0.02 mM, NS) or HbA1c level (4.42 +/- 0.2 vs. 4.6 +/- 0.3%, NS) between gestational and nondiabetic patients. Diabetic patients were followed at 2-wk intervals, with serum glycosylated protein analysis, HbA1c, fasting glucose, and mean glucose determined by outpatient monitoring. Serum glycosylated protein correlated significantly to fasting blood glucose (r = 0.81, P less than 0.001) and mean outpatient glucose (r = 0.62, P less than 0.001) at the 2-wk follow-up visits. No correlation was found between HbA1c and fasting blood glucose (r = 0.11, NS) or mean outpatient glucose (r = -0.12, NS) during the follow-up period. The serum glycosylated protein level (serum fructosamine) is not a useful screening test for gestational diabetes. However, this assay shows potential as an objective marker of short-term control in evaluating the maternal glycemic state.