The pathogenesis of normal-tension glaucoma remains unknown. Because ocular vasospasm has been proposed as a possible mechanism, we investigated ocular vessel flow velocity in normal-tension glaucoma patients at rest and under treatment with a cerebral vasodilator. Ten normal-tension glaucoma patients and nine age- and gender-matched controls had flow velocity measured in three vessels (ophthalmic artery, central retinal artery, and temporal short posterior ciliary artery) by using color Doppler imaging, under baseline conditions and during carbon dioxide supplementation sufficient to increase end-tidal PCO2 by 15%. Peak systolic and end-diastolic velocities were measured, and the resistance index (peak systolic velocity minus end-diastolic velocity, divided by peak systolic velocity) was calculated. Compared with controls, these normal-tension glaucoma patients had significantly lower end-diastolic velocities (P = .002) and higher resistance indices (P = .007) in the ophthalmic artery at baseline. When PCO2 was increased, control subjects remained unchanged, whereas it increased end-diastolic velocity in patients (P = .003) and abolished the difference in resistance index between the two groups. Patients and control subjects differed little in their baseline or carbon dioxide response velocities or in resistance in the other two vessels. These results indicate that at baseline these normal-tension glaucoma patients may have increased vascular resistance distal to the ophthalmic artery, although this increased resistance cannot be specifically ascribed to the central retinal arterial or to temporal short posterior ciliary arterial vascular beds. The responsiveness of these patients to a cerebral vasodilator (increased PCO2) indicates further that the increased resistance distal to the ophthalmic artery may be the reversible result of vasospasm.