The glycation gap (G-gap: difference between measured hemoglobin A1c [A1C] and the value predicted by its regression on the fructosamine level) is stable and associated with diabetic complications. Measuring A1C level in International Federation of Clinical Chemistry (IFCC) units (A1C-SI; mmol/mol) and National Glycohemoglobin Standardization Program units (A1C-NGSP; %) and using glycated albumin (GA) level instead of fructosamine level for calculating the G-gap, we investigated whether the G-gap is better represented by GA/A1C ratio if expressed in SI units (GA/A1C-SI ratio) rather than in NGSP units (GA/A1C-% ratio). We examined 749 Japanese children with type 1 diabetes using simultaneous GA and A1C measurements. Of these, 369 patients were examined more than five times to assess the consistency of the G-gap and the GA/A1C ratio within individuals. The relationship of GA/A1C-% ratio to the corresponding A1C-NGSP was stronger than that of GA/A1C-SI ratio to A1C-IFCC. At enrollment, the inverse relationship between the GA/A1C-SI ratio and G-gap was highly significant (R(2) = 0.95) compared with that between the GA/A1C-% ratio and G-gap (R(2) = 0.69). A highly significant inverse relationship was also observed between the mean GA/A1C-SI ratio and the mean G-gaps obtained individually over time (R(2) = 0.95) compared with that using the corresponding A1C-NGSP (R(2) = 0.67). We conclude that the G-gap is better represented by the GA/A1C-SI ratio. We propose the use of mean GA/A1C-SI ratios easily obtained individually over time as reference values in Japanese children with type 1 diabetes (6.75 ± 0.60 [means ± SD]).