PURPOSE: To determine pressure-flow characteristics at physiologic flow rates in vitro and in vivo in rabbits for Ahmed, Baerveldt, Krupin disk, and OptiMed glaucoma implants. The Molteno dual-chamber implant also was evaluated in vivo only. METHODS: Five samples of each glaucoma implant were studied. Baerveldt implants were ligated partially for in vitro testing. Opening and closing pressures in air or after immersion in balanced salt solution or plasma were evaluated for the valved devices (Ahmed and Krupin). Pressures were measured in vitro and in vivo in normal rabbits at flow rates preset at between 2 and 25 microliters/minute after the tubes were connected to a closed manometric system. In vivo measurements were made 24 hours after implantation. Resistance to flow was calculated using Poiseuille's equation after at least three separate flow rate readings. RESULTS: In air, the Ahmed and Krupin valves had opening pressures of 9.2 +/- 3.4 and 7.2 +/- 0.6 mmHg and closing pressures of 5.2 +/- 0.9 and 3.9 +/- 1 mmHg, respectively. Neither opening nor closing pressures could be determined when Ahmed and Krupin valves were immersed. In vitro, the Ahmed and OptiMed devices had higher pressures than did other devices at a 2-microliters/minute flow rate of balanced salt solution. During perfusion with plasma, only the OptiMed device maintained higher pressures than with balanced salt. With all devices, pressures fell rapidly to zero after flow was stopped. The OptiMed device demonstrated the highest resistance values. In vivo, the Ahmed device provided pressures of 7.5 +/- 0.8 mmHg and the OptiMed device gave pressures of 19.6 +/- 5.6 mmHg at a 2-microliters/minute flow rate. After 15 minutes of flow shutdown, the OptiMed implant maintained pressures of 7.1 +/- 1.1 mmHg. The Baerveldt (nonligatured), Krupin, and Molteno dual-chamber implants had similar resistances and pressures in vivo. Pressures with all devices in vivo fell rapidly to zero after conjunctival wound disruption. CONCLUSION: Neither the Ahmed nor Krupin devices had demonstrable opening or closing pressures when tested in vitro immersed in balanced salt solution or plasma. With all devices, pressures were higher in vivo than in vitro due to tissue-induced resistance around the explant. Both Ahmed and Krupin valves functioned as flow-restricting devices at the flow rates studied, but did not close after initial perfusion with fluid.