Services for children with diabetes are a good model for audit because of relatively well-defined and available measures of outcome. Estimation of glycated haemoglobin (HbA1) is accepted as one index of control, although the assay and normal ranges vary in different centres. Sheffield has a stable clinic population and we have taken blood for HbA1 on each patient two to four times per year since 1983. HbA1 values have been aligned in 3-monthly increments from time of diagnosis to construct a graph showing the mean plus or minus two standard deviation values for the clinic population and also a rough 'centile chart' that describes the unfortunate but expected pattern of deteriorating control with duration of diabetes. There are undoubtedly methodological flaws in this novel approach; however, such charts constructed for each clinic may be a suitable basis for comparative assessment of the control of an individual and also for auditing the effects of changes in practice that strive for optimum glucose levels in the clinic population.