The relationship between the progression of diabetic retinopathy and consecutive blood glucose control evaluated by glycosylated hemoglobin (HbA1) values was statistically analyzed based on 422 eyes of 211 diabetic subjects who were followed up 2 years or more. The Cox regression model (proportional hazard model) with HbA1, as a time-dependent covariate was applied in the analysis. The cumulative effect and the relative risk of HbA1 affecting the progression of diabetic retinopathy was estimated. The effect of the moving average of HbA1 values for the previous 1-30 months on the progression of retinopathy was assessed to investigate the cumulative effect of HbA1. The cumulative effect of HbA1 on the progression of diabetic retinopathy was significant when the time span of the moving average was 6 months or more than 6 months. The moving average of HbA1 value for 7 months was most relevant to the retinopathy progression. The relative risk corresponding to the increase in the moving average of HbA1 values for 6 months by 1, 2 or 3% was 1.18, 1.39 or 1.63, respectively. When the HbA1 level was adjusted to its average value in the subjects (9.6%), the estimated cumulative retinopathy progression rate for 6, 12, 18 or 24 month follow-up was 3.6, 14.5, 22.5 or 30.5%, respectively. The differences in the cumulative retinopathy progression rate between a well-controlled group with a low moving average of HbA1 value (3% lower than the average) and a poorly-controlled group with a high moving average of HbA1 value (3% higher than the average) were estimated as 3% (6-month follow-up) and 23% (24 months).