This cohort study investigated the association between treatment cessation and incidence/progression of diabetic retinopathy (DR) in Japanese patients with type 2 diabetes mellitus (T2DM). Data were extracted from electronic medical records at the University of the Ryukyu Hospital and the Tomishiro Central Hospital of Okinawa, Japan. We enrolled 417 diabetic patients without DR (N = 281) and with nonproliferative DR (N = 136) at the baseline. Treatment cessation was defined as failing to attend outpatient clinics for at least twelve months prior to the baseline. After a median follow-up of 7years, we compared the incidence/progression rate of DR including nonproliferative and proliferative DR between patients with and without treatment cessation and calculated the odds ratio (OR) in the treatment cessation group using a logistic regression model. The overall prevalence of treatment cessation was 13% in patients with T2DM. Characteristics of treatment cessation included relative youth (57 ± 11years vs. 63 ± 12years, P < 0.01). Treatment cessation was tightly associated with the incidence of DR (OR 4.20 [95% confidence interval [CI] 1.46-12.04, P < 0.01) and also incidence/progression of DR (OR 2.70 [1.28-5.69], P < 0.01), even after adjusting for age, sex, BMI, duration of T2DM, and HbA1c level. By considering major confounding factors, the present study demonstrates an independent association between treatment cessation and incidence of DR in patients with T2DM, highlighting treatment cessation as an independent risk for DR in T2DM. The online version contains supplementary material available at 10.1007/s13340-024-00724-7.