The potential hemostatic effect of an intravitreal air bubble after diabetic vitrectomy was studied in an animal model and in a randomized clinical trial. One day after vitrectomy with induced intraoperative hemorrhage, vitreous cavity hemorrhage was present in 60% of air-filled rabbit eyes compared with 27% of fluid-filled eyes. The prevalence and extent of hemorrhage was equal in the two groups on postoperative days 3 and 7. In a clinical trial of 51 eyes undergoing diabetic vitrectomy, 70% of eyes randomized to air-filled vitreous cavity after vitrectomy had vitreous cavity hemorrhage on postoperative day 1 compared with 50% of fluid-filled eyes. At 1 week, the incidence of hemorrhage was 78% for air and 61% for fluid. The 6-month visual and anatomic results were similar in both groups. These findings suggest that an intravitreal air bubble neither improves hemostasis nor reduces the visual outcome after diabetic vitrectomy.