Objective: To approach the possibility of evaluating perinatal cardiac function in fetuses of diabetic mothers by measuring brain natriuretic peptide (BNP) in umbilical cord plasma. Methods: 24 gestational diabetic mothers were divided into good and poor control groups due to their blood glucose before delivery. Fetal cardiac sizes and function were measured in their late pregnant periods by fetal echocardiography. 25 uncomplicated mothers were as control. Cord bloods were sampled for detecting the BNP concentrations. Results: Among the groups the fetuses’ gestational weeks and weights at birth had no significant differences, and the BNP concentrations had no correlations with them. The BNP concentrations in diabetic group were significantly higher than control (114.0±39.0 vs. 80.6±13.7 pg/ml, P<0.001), among which the poor control group higher than the good control group (142.1±44.1 vs. 102.4±31.2 pg/ml, P<0.01, n=7 and 17 respectively). No significant difference was found between the gestational diabetes mellitus and impaired glucose tolerance mothers, and no correlation of the BNP concentrations to these mothers’ blood glucose at diagnosis. The BNP concentrations positively correlated to the fetal left ventricular walls thicknesses and the peak velocities of mitral A wave, and negatively correlated to the mitral E/A ratios (r =0.715, 0.491 and -0.507 respectively,P<0.05). Conclusions: BNP in cord blood is possible to evaluate perinatal cardiac function. diabetic pregnant fetuses have an increased BNP level in cord blood, which accords with maternal glucose control and changes in fetal diastolic function. Good glucose control may decrease the impact of abnormal maternal blood glucose on fetal hearts.