Background As one of the most common pregnant complications, the gestational diabetes mellitus (GDM) is associated with significant adverse pregnant outcomes and it is crucial to accurately monitor the glycemic states of GDM patients. The HbA1c which is a traditional long-term glycemic marker used in diabetic patients, is not recommended in GDM patients during pregnancy. Recently, many efforts have been focused on the alternative marker glycated albumin (GA) and its application in pregnancy during which profound physiological changes take place. Our objective was to determine the reference intervals (RIs) of GA in healthy Chinese pregnant women and to assess the predictive value of serum GA in adverse pregnant outcomes. Methods Totally 479 healthy subjects including 153 in the first trimester, 174 in the second trimester, and 152 in the third trimester were enrolled from March to July 2019, for the purpose of establishing the trimester-specific RIs of GA. The diagnostic value of GA for GDM patients was evaluated and compared with that of fasting plasma glucose (FPG) at 24-28 weeks of gestation. The association between GA in the late pregnancy and the adverse pregnant outcomes was analyzed retrospectively with the data collected from January to June 2018 at our hospital. Results The estimated RIs of GA in present study were 10.87-15.09 %, 10.04-13.50 %, and 9.78-13.03 % in the first, second, and third trimesters respectively. The areas under receiver operating characteristic (ROC) curves were 0.503 for GA and 0.705 for FPG. More importantly, the GA levels of the third trimester did not show significant changes in women with large-for-date birth weight, preterm delivery, postpartum hemorrhage or hypertension when compared in women with normal pregnancy outcomes. The exception was that the GDM patients who developed preeclampsia did have a lower GA level in their late pregnancy. Conclusions Our results show that the GA was continuously decreased as the gestational age went up. It has limited value in diagnosing GDM and predicting adverse pregnancy outcomes.