AIM: To determine whether glycated haemoglobin (HbA(1c)) can be used in combination with fasting plasma glucose (FPG) for the diagnosis of diabetes in patients with impaired fasting glucose (IFG) and in a broader spectrum of patients. METHODS: An algorithm was derived from oral glucose tolerance test (OGTT) capillary samples in 500 consecutive UK patients with IFG by World Health Organization criteria. It was validated in a further 500 UK patients and, with venous specimens, in 1175 unselected Australian patients. RESULTS: The derivation cohort was aged 61 years (50-69 years) (median IQ range) with 52% male and 12% South Asian. Diabetes Control and Complications Trial-aligned HbA(1c) was 6.2% (5.8-6.6%) (reference interval <6.0%) and FPG 6.7 mmol/l (6.3-7.2 mmol/l). FPG was in the diabetes range in 36% of patients, with an OGTT identifying a further 12% with diabetes. The derived algorithm, (HbA(1c) >or= 6.0% with FPG <7.0 mmol/l) identified those patients requiring an OGTT to diagnose diabetes. When applied to the UK validation cohort, sensitivity was 97% and specificity 100%. The algorithm was equally effective in the unselected group, aged 59 years (49-68 years) and 54% male, with sensitivity 93% and specificity 100%. HbA(1c) was 6.0% (5.6-6.6%) and FPG 6.0 mmol/l (5.3-6.8 mmol/l), with 26% having IFG. Use of the algorithm would reduce the number of OGTTs performed in the UK validation cohort by 33% and by 66% in the Australian patients studied. CONCLUSIONS: Use of this algorithm would simplify procedures for diagnosis of diabetes and could also be used for monitoring pre-diabetes. Validation is now required in other populations and patient groups.