PURPOSE: To assess the visual outcome following vitrectomy for diabetic retinopathy compared with previous studies; to evaluate risk factors for light perception (LP) and no light perception (NLP) vision after diabetic vitrectomy. DESIGN: Retrospective medical record review. METHODS: The charts of 100 consecutive patients who underwent vitrectomy for proliferative diabetic retinopathy between November 1, 1997, and November 30, 1998, were reviewed. Surgical technique included standard pars plana vitrectomy with combination of delamination and segmentation of gliotic tractional membranes using bimanual techniques. All patients had post-operative follow-up of at least 12 months. Several factors were analyzed for their effect on poor visual outcome (LP and NLP) using Fisher's exact test. RESULTS: Post-vitrectomy, 73% of diabetic patients had stable or improved vision; 16% had worsened but functional vision, defined as worse but still < 20/400; 4% had worsened but ambulatory vision, defined as worse but still count fingers (CF) or hand motion (HM); and 7% had poor visual outcome, LP or NLP. Resultant visual acuity was <20/40 in 38% of patients, 20/50 to 20/100 in 34%, 20/120 to CF in 18%, HM in 3%, LP in 4%, and NLP in 3%. Risk factors for eyes with LP and NLP vision included pre-operative iris neovascularization (INV), P = .05, post-operative INV, P = .02, post-operative macular ischemia, P = .0001, and post-operative vitreous hemorrhage (VH), P = .02. CONCLUSIONS: Pre-operative and post-operative INV, post-operative macular ischemia, and post-operative VH appear to be risk factors for LP and NLP vision following diabetic vitrectomy, whereas overall improvements in surgical technique and visual outcome continue to be reported. © 2005 by Elsevier Inc. All rights reserved.