To investigate the effects of vitrectomy combined with internal limiting membrane (ILM) peeling indiabetic macular edema (DME). Methods: In this retrospective clinical study, 33 eyes of 31 patients with DME confirmed preoperatively or intraoperatively by optical coherence tomography (OCT) were included. All patients underwent vitrectomy for dense vitreous hemorrhages due to diabetic retinopathy or proliferative diabetic retinopathy between June 2014 and January 2017 at the Eye Hospital,Wenzhou Medical University. A total of 16 patients (18 eyes) with vitrectomy and ILM peeling were in the ILM-removed group, 15 eyes of 15 patients with vitrectomy only were in the control group. All the surgeries were performed by the same surgeon. All subjects underwent OCT examination at 1 month and 3 months postoperatively. The central macular thickness (CMT) and visual outcomes between the two groups were analyzed relative to each other. The data were analyzed by repeated measures analysis and t-tests. Results: Before the therapy, and at one and three months after therapy, the total difference of best corrected visual acuity (BCVA) between the two groups was statistically significant (F=15.93, P<0.001). The BCVA in the ILM-removed group was higher than the control group at one month after therapy (t=2.55, P=0.02). However, there was no significant difference between the two groups in terms of BCVA at three months after therapy (t=0.82, P=0.42). Before the therapy, and at one and three months after therapy, the total difference of CMT between the two groups was not statistically significant (F=2.85, P=0.065). At both one month and three months after therapy, the CMT in the ILM-removed group was lower than the control group (t=2.24, P=0.03; t=3.79, P=0.001). At 1 month postoperatively, the subjects of effectiveness (a decrease in CMT by at least 20%), ineffectiveness (a change in CMT by <20%) and deterioration (an increase in CMT by more than 20%) were 8, 6 and 4 respectively, while the subjects were 11, 5 and 2 respectively at 3 months after therapy. There was no significant difference from the control group (Z=-1.687, P=0.092) at one month after therapy, but there was a significant difference at three months after therapy (Z=-2.177, P=0.029). Conclusions: The removal of ILM contributes to the resolution of nontractional DME at an early stage after therapy.