The authors determined the anatomic and visual success rates of vitrectomy in patients who have failed previous macular hole surgery. Standardized Early Treatment Diabetic Retinopathy Study (ETDRS)-style best-corrected visual acuity was measured in each patient pre- and postoperatively. Surgery was performed at two centers using a standardized protocol, as previously reported, using 1330 ng bovine-derived transforming growth factor-beta 2 (TGF-beta 2) and 16% perfluoropropane internal gas tamponade. Endpoint analyses were conducted for anatomic closure of the hole, visual improvement of three or more ETDRS lines, and final visual acuity of > or = 20/63 and > or = 20/40. Forty-eight failing macular hole surgery were reoperated. The anatomic results showed closure of the hole in 40 (83%) cases at the end of the mean follow-up interval of 7.4 months. The visual acuity improved in greater than or equal to three lines in 52%, was > or = 20/63 in 54%, and was > or = 20/40 in 25% of reoperated eyes. The only factor associated with a better final visual acuity was preoperative visual acuity better than 20/80. Nuclear sclerosis leading to cataract extraction was observed in 13 (30%) of 43 initially phakic eyes. Macular hole closure was induced in this subset of reoperated patients. Visual improvement occurred often, although in a lower percentage than has been reported for primary surgical eyes. Repeat vitrectomy should be considered in patients with persistent macular hole after failure of primary surgery.