The article discusses the comparison of two methods for measuring glycated haemoglobin (HbA1) – electrophoresis and HbA1c. It highlights that when comparing results from similar methods, differences in the proportion of patients classified as having poorly controlled diabetes may disappear. The 5 SD limit for HbA1c is 5.44%, representing an increase of 1.35 times the mean concentration in the non-diabetic reference population. For HbA1 measured by electrophoresis, this corresponds to 8.51% or 2.95 SD. The proportion of the population within this limit is estimated to be 0.25, similar to that for HbA1c. However, the total relative frequency for HbA1 measured by electrophoresis is not 1.0, so this estimate needs confirmation. The authors suggest that comparisons should be based on multiples of a measure of central location (mean, median, or mode) rather than SD. Using multiples of the mean for HbA1c, the cutoff points are approximately 1.20 and 1.35 times the mean. If data were reanalysed and like was compared with like, the reported difference in the proportion of patients categorised as having poorly controlled disease would probably disappear. This would lead to different conclusions, with points three to five of the paper's clinical implications needing to be withdrawn. The authors also argue that using SD is inappropriate for comparing methods of measuring glycated haemoglobin when coefficients of variation in the control population differ between the methods. This method of comparison was used in the European IDDM Policy Group guidelines, which should be reconsidered. In response to a comment, the authors agree that using a fixed biological variation to express SD may help in comparing methods of measuring HbA1 with different imprecisions. They suggest expressing a diabetic patient's result as a multiple of the mean value in non-diabetic people (MoM). This method may be valid for comparing two methods of measuring HbA1 or HbA1c, but comparisons between HbA1 and HbA1c remain problematic. Standardisation in measurements of glycated haemoglobin is needed.The article discusses the comparison of two methods for measuring glycated haemoglobin (HbA1) – electrophoresis and HbA1c. It highlights that when comparing results from similar methods, differences in the proportion of patients classified as having poorly controlled diabetes may disappear. The 5 SD limit for HbA1c is 5.44%, representing an increase of 1.35 times the mean concentration in the non-diabetic reference population. For HbA1 measured by electrophoresis, this corresponds to 8.51% or 2.95 SD. The proportion of the population within this limit is estimated to be 0.25, similar to that for HbA1c. However, the total relative frequency for HbA1 measured by electrophoresis is not 1.0, so this estimate needs confirmation. The authors suggest that comparisons should be based on multiples of a measure of central location (mean, median, or mode) rather than SD. Using multiples of the mean for HbA1c, the cutoff points are approximately 1.20 and 1.35 times the mean. If data were reanalysed and like was compared with like, the reported difference in the proportion of patients categorised as having poorly controlled disease would probably disappear. This would lead to different conclusions, with points three to five of the paper's clinical implications needing to be withdrawn. The authors also argue that using SD is inappropriate for comparing methods of measuring glycated haemoglobin when coefficients of variation in the control population differ between the methods. This method of comparison was used in the European IDDM Policy Group guidelines, which should be reconsidered. In response to a comment, the authors agree that using a fixed biological variation to express SD may help in comparing methods of measuring HbA1 with different imprecisions. They suggest expressing a diabetic patient's result as a multiple of the mean value in non-diabetic people (MoM). This method may be valid for comparing two methods of measuring HbA1 or HbA1c, but comparisons between HbA1 and HbA1c remain problematic. Standardisation in measurements of glycated haemoglobin is needed.