Objectives. To investigate the ability of classifying diabetes mellitus with clinical and biochemical characteristics at diagnosis in 25᎐45 years old patients. Design. We determined age, body mass index (BMI), ICA, fasting C-peptide and HbA 1c at diagnosis, and after 4᎐27 months follow up, these variables were related to the type of treatment. Setting. Omdurman Teaching Hospital, Sudan. Subjects. Eighty-six consecutive newly diagnosed diabetic patients, according to WHO criteria, were included. Results. At diagnosis, 27 patients (31.4%) were treated with insulin and 59 (68.6%) with diet and/or oral hypoglycaemic agents (OHA). Insulin-treated patients at diagnosis were younger, had lower BMI and higher HbA 1c . In the non-insulin treated group at diagnosis, ICA-negative patients (n ϭ 54) had higher fasting Cpeptide (P Ͻ 0.05) than ICA-positive. The treatment was changed from insulin to diet and/or OHA in 17 patients, and to insulin in six patients. Patients who discontinued insulin had already at diagnosis higher BMI and higher fasting C-peptide concentration (P Ͻ 0.005 and P Ͻ 0.05, respectively) than patients remained on insulin. The positive predictive values for insulin treatment of ICA positivity, low fasting Cpeptide, and low BMI were 90, 78 and 73%, respectively. The sensitivity, specificity and positive predictive values for these variables in concurrence were, respectively, 85, 92 and 69%. Conclusion. The clinical classification of newlydiagnosed young Sudanese patients was satisfactory. With regular follow up and early commencement of appropriate treatment, good glycaemic control is certainly attainable. BMI is potentially valuable for the clinical classification in such patients. ICA and fasting serum C-peptide facilitate the choice of insulin treatment, but cost is a limit to large-scale use of these tests.