PURPOSE: To identify predictive factors that help determine the interval of intravitreal anti-vascular endothelial growth factor (anti-VEGF) injection after the initial resolution of diabetic macular edema (DME). METHODS: This retrospective case-control study enrolled treatment-naïve DME patients who had achieved DME resolution after intravitreal anti-VEGF injections. Patients were classified into the recurrence and no-recurrence groups, depending on the development of recurrent DME after deferring intravitreal anti-VEGF injection. The demographics and clinical features, including optical coherence tomography findings, were compared between the two groups. RESULTS: We enrolled 105 eyes. Sixty eyes (57.1%) belonged to the no-recurrence group, and 45 (42.9%), belonged to the recurrence group. The severity of diabetic retinopathy at baseline was related to early DME recurrence (P = 0.009). At the treatment deferring point, the non-recurrence group had both thinner central subfield thickness (289.5 ± 27.2 μm vs. 307.0 ± 38.2 μm, P = 0.011) and thinner central retinal thickness (214.9 ± 41.4 μm vs. 231.8 ± 41.2 μm, P = 0.043) compared to the recurrence group. Intraretinal cyst was observed in 34 eyes (56.7%) in the no-recurrence group and 42 eyes (93.3%) in the recurrence group at the deferring point (P < 0.001). CONCLUSION: A low risk of early DME recurrence is anticipated in the eyes with foveal thinning and no intraretinal cyst when anti-VEGF injection is deferred. These predictive biomarkers can be useful for patient monitoring and determining treatment strategies for DME patients.