This study evaluates the effects of glucagon 30 mug/kg (maximal 1 mg) on beta-cell function in children by C-peptide determined before and 6 min after intravenous administration. From 1990 to 2005, 118 Taiwanese children with newly diagnosed diabetes mellitus (98 children with type 1 and 20 children with type 2) and 29 normal Taiwanese children were enrolled in this study. Fasting and 6-min post-glucagon C-peptide levels were analyzed. In the pre-pubertal group, the median fasting serum C-peptide levels were 0.2 and 0.8 nmol/l in type 1 diabetes and normal children, respectively. These levels rose to 0.3 and 1.9 nmol/l after glucagon stimulation. In the pubertal group, the median fasting serum C-peptide levels were 0.3, 1.0 and 0.9 nmol/l in type 1 diabetes, type 2 diabetes and normal children, respectively. They rose to 0.4, 2.5 and 2.7 nmol/l after glucagon stimulation. Both fasting and post-glucagon C-peptide levels in type 1 diabetes patients were significantly lower than those of normal children and children with type 2 diabetes. The optimal cut-off values to distinguish type 1 diabetes patients from those with type 2 as determined by the receiving operating characteristic curve were 0.7 and 1.1 nmol/l, respectively. The sensitivities of both C-peptide values were 93%. The post-glucagon C-peptide level was more powerful in distinguishing type 1 diabetes from type 2 diabetes with higher specificity (95% vs. 85%). The 6-min glucagon test is valuable in assessing beta-cell function in children and can help pediatricians in the differential diagnoses of diabetes mellitus in children.