To verify whether hemoglobin A1c (HbA1c) levels in early pregnancy can predict the diagnosis of gestational diabetes mellitus (GDM) in mid-pregnancy. This was a retrospective cohort study of 2008 pregnant women who delivered singletons at the Yokohama City university Medical Center. Concomitant or history of diabetes mellitus and overt diabetes in pregnancy were excluded. Pregnant women at high risk for GDM underwent a one-step 75-g oral glucose tolerance test (OGTT) during mid-pregnancy. For other pregnant women, GDM was diagnosed by a two-step 75-g oral glucose tolerance test (OGTT) when the 50-g glucose challenge test result in mid-pregnancy was ≥140 mg/dL. The thresholds for 75-g OGTT followed those of the International Association of Diabetes and Pregnancy Study Group (IADPSG) criteria (92-180-153 mg/dL). The relationship between HbA1c level measured at <20 weeks of gestation and GDM diagnosis at mid pregnancy was assessed using a receiver operating characteristic curve (ROC); area under the curve (AUC) and optimal cutoff value of HbA1c, predictive of GDM were calculated. The median HbA1c level at <20 weeks of gestation was 5.3%, and 8.5% of women were diagnosed with GDM. In the ROC curve of the GDM diagnosis rate by HbA1c level, AUC was 0.706, and the optimal cutoff value was 5.4%, with a sensitivity of 0.6176, specificity of 0.6834, positive predictive value of 15.4%, and negative predictive value of 95.1%. Although HbA1c at less than 20 weeks of gestation is acceptable discrimination as a diagnostic tool of GDM in mid-pregnancy, it is not clinically useful to predict GDM in mid-pregnancy.