To prospectively assess the risks and benefits of vitrectomy surgery for eyes with stage 3 or 4 macular holes. A multicentered, controlled, randomized clinical trial. Community- and university-based ophthalmology clinics. One hundred twenty patients (129 eyes) with stage 3 or 4 macular holes. Standardized macular hole surgery vs observation alone. Four measures of best-corrected visual function, standardized photographic evaluation of the extent of hole closure, evaluation of lens opacification, and determination of adverse events. Outcomes were determined at 6 months after randomization. Compared with observation alone, a significant benefit due to surgery was found in the rate of hole closure (4% vs 69%, P < .001). After adjusting for baseline visual acuity, hole duration, and maximum hole diameter, a significant benefit due to surgery was found in visual acuity for the Bailey-Lovie Word Reading (P = .02) and the Potential Acuity Meter (P < .01) tests; a marginally significant benefit due to surgery was found in visual acuity for the Early Treatment Diabetic Retinopathy Study chart (P = .05). Although the proportion of eyes achieving a change in visual acuity of 2 or more lines on the Early Treatment Diabetic Retinopathy Study chart was significantly greater for the surgery group vs the observed group (11 [19%] of 59 eyes vs 3 [5%] of 58 eyes, adjusted P = .05), 20 (34%) of 59 eyes randomized to surgery had a loss in visual acuity of 1 or more lines. Compared with the observation group, eyes randomized to surgery had higher nuclear sclerosis scores (2.4 vs 1.3, P < .001). Fourteen adverse events were noted in the surgery group; none were noted in the observed group. Some visual benefit of vitrectomy surgery for macular holes exists, despite a notable incidence of adverse events. The large variability in visual acuity outcome in the surgical group may be because of complications or progressive cataract. A study of the long-term outcome after macular hole surgery is needed.