This study compared glycosylated hemoglobin (HbA1c) levels in the first and second trimesters of pregnancy and assessed maternal and neonatal outcomes according to HbA1c variations among women with pregestational diabetes. This retrospective, multicenter Korean study involved mothers with diabetes who had given birth in 17 hospitals. A total of 292 women were divided into three groups based on HbA1c levels during the first and second trimesters: women with HbA1c levels maintained at <6.5% (well-controlled [WC] group); women with HbA1c ≥ 6.5% (poorly-controlled [PC] group); and women with HbA1c ≥ 6.5% in the first trimester but <6.5% in the second trimester (improved-control [IC] group). The PC group had the highest pregnancy-associated hypertension (PAH) incidence, while the incidence did not significantly differ between the WC and IC groups. The receiver operating characteristic (ROC) curve indicated that HbA1c in the second trimester could predict PAH with a cut-off value of 5.7%. The PC versus WC versus IC group showed statistically significantly higher neonatal birthweight and significantly higher rates of large for gestational age (LGA); however, those were not significantly different between the WC and IC groups. HbA1c levels in the second trimester could predict LGA, with a cut-off value of 5.4%. Therefore, the second trimester HbA1c levels were significantly associated with both maternal and neonatal outcomes.