The safety and success rates of trabeculectomy using fornix-based or limbus-based conjunctival flaps were investigated in a randomized, prospective study. Patients with phakic eyes and uncontrolled primary open-angle glaucoma had a trabeculectomy with either a standard limbus-based conjunctival flap (18 eyes) or a fornix-based conjunctival flap (19 eyes). Filtration surgery was equally successful (intraocular pressure, less than or equal to 21 mm Hg) in both groups independent of the type of conjunctival flap: 17 (94%) of 18 eyes with a limbus-based flap (three eyes with additional medical therapy) and 17 (89%) of 19 eyes with a fornix-based flap (three eyes with additional medical therapy). Four eyes in the fornix-based group showed positive Seidel tests in the early postoperative period. Aqueous leakage was from the suture sites at the lateral margins of the flap and resolved spontaneously without surgical intervention. The fornix-based conjunctival flap was easier to perform, provided better surgical exposure, and was easier to close than the limbus-based flap.