To compare choroidal blood pressure versus flow relationships obtained by three different methods of changing the ocular perfusion pressure. Experiments were performed in pentobarbital-anesthetized rabbits with occluders on the aorta and inferior vena cava to control mean arterial pressure (MAP). The central ear artery was cannulated to measure MAP. Two 23-gauge needles were inserted through the pars plana into the vitreous: one connected to a saline-filled syringe to vary the ocular volume and the other to a pressure transducer to measure intraocular pressure (IOP). Choroidal perfusion was measured by laser-Doppler flowmetry with the probe in the vitreous over the posterior pole. In group 1 (n= 15), the MAP was varied while holding the IOP at 10, 15, 20, 25 and 30 mm Hg. In group 2 (n= 19), the IOP was increased while holding the MAP at 80, 70, 60, 50, 40, 30 and 20 mm Hg. In group 3 (n= 21), the MAP was varied without controlling the IOP. Group 1 baseline choroidal flows were similar at the five IOPs. When the flow was plotted against MAP, the curves diverged and extrapolated to intersect the pressure axis when the MAP equaled the set IOP. Group 2 baseline flows were similar at MAPs greater than 40 mm Hg. When the flow was plotted against the IOP, the curves diverged and intersected the pressure axis when the IOP equaled the MAP. In both groups, plotting the flow against the perfusion pressure (ie, MAP minus IOP) collapsed the data points into single curves. Choroidal autoregulation occurred in all three groups; however, the low end of the autoregulatory perfusion pressure range was approximately 50 mm Hg in group 1, approximately 40 mm Hg in group 2, and approximately 30 mm Hg in group 3. The results show that the effective choroidal perfusion pressure gradient equals the MAP minus the IOP, and that choroidal autoregulation is most effective when the MAP varies and IOP is not controlled.