BACKGROUND Clinical criteria and the presence of auto-antibodies might not be sufficient to discriminate type 1 diabetes mellitus (T1DM) and type 2 diabetes mellitus (T2DM) in children with new onset of disease. OBJECTIVE To determine whether biochemical markers can be used to distinguish T1DM from T2DM. DESIGN AND INTERVENTION This was a single-center, US-based, prospective study of children with newly diagnosed primary diabetes. Children with secondary diabetes as a result of medication, cystic fibrosis, or other medical conditions were excluded. Eligible participants were enrolled between July 1999 and July 2000. T1DM was presumptively diagnosed in nonobese children who were dependent on insulin to maintain glycemic control. T2DM was presumptively diagnosed in children with a BMI >85% for age and sex and a family history of T2DM, who were insulin-independent and glutamic acid decarboxylase autoantibody-negative. Glucose, CO2, urinary ketones, and HbA1c levels were measured at diagnosis. A fasting blood sample was collected on the morning of diagnosis or on the morning after diagnosis to determine the levels of C-peptide and insulin-like-growth-factor-binding protein 1 (IGFBP-1). C-peptide was assessed by immuno-chemiluminescent assay and IGFBP-1 was assessed by radioimmunoassay. OUTCOME MEASURES The main outcome measures were the receiver operator characteristic (ROC), area under the curve (AUC), sensitivity, specificity, and cut-off value of each biochemical marker. RESULTS The study enrolled a total of 175 children, 149 of whom were presumptively diagnosed with T1DM. The mean age of children with T1DM was 8.7 years, compared with a mean age of 14.2 years for children with T2DM (P<0.001). Children with T1DM were most likely be non- Hispanic whites (78%, P<0.001). By contrast, children with T2DM were most likely to be non- Hispanic blacks (58%, P<0.001). Markers of insulin resistance demonstrated a significant difference between T1DM and T2DM. The mean fasting C-peptide level was 0.13 nmol/l in T1DM and 0.88nmol/l in T2DM (P<0.001). The ROC curve for fasting C-peptide determined a cut-off value of 0.28nmol/l, with a sensitivity of 83%, a specificity of 89%, and an AUC of 0.94. The mean fasting IGFBP-1 level was 38.1 ng/ml in T1DM and 3.6ng/ml in T2DM (P<0.001). The ROC curve for fasting IGFBP-1 determined a cut-off value of 3.6ng/ml, with a sensitivity of 93%, a specificity of 79%, and an AUC of 0.92. The CO2 level for children with TIDM was 17.9mmol/l, versus 22.7mmol/l for children with T2DM (P= 0.016). The ROC curve for CO2 determined a cut-off value of 21.5 mmol/l, with a sensitivity of 83%, a specificity of 64%, and an AUC of 0.76. Urinary ketones were present in 79% and 56% of children with T1DM and T2DM, respectively. A cut-off value of 10 mg/dl was associated with a sensitivity of 74%, a specificity of 52%, and an AUC of 0.73. CONCLUSION Children with new-onset T1DM and T2DM can be distinguished by a combination of biochemical markers. [ABSTRACT FROM AUTHOR]