The fasting plasma glucose assay was compared with the one-hour post-glucose test as a screening test for identification of gestational diabetes. Of 4,561 consecutive patients screened with a 50-g glucose test, 968 (21.2%) had results > or = 135 mg/dL; 141 (14.6%, or 3.1% of the total) were found to have diabetes. In the 968 patients, the area under the fasting plasma glucose receiver operating characteristic curve was greater than that under the glucose screening test curve, indicating greater discriminatory value of the former test. Of the 116 patients who had sequential glucose screening tests and fasting plasma glucose assays performed twice during pregnancy, a significant correlation was found for fasting plasma glucose values, but not for glucose screening test values. We conclude that the fasting plasma glucose assay may perform better than the one-hour post-glucose test as a screening test for gestational diabetes. Based on these data, a population-based prospective study seems justified.