Measurements of pupillary reactivity and size were recorded using neutral density filters and infrared pupillometry (IRP) in a prospective masked study of acute central retinal vein occlusion (CRVO) to quantify the two methods of measurement and to compare their value in the prediction of rubeosis. Thirty two patients were examined within 45 days of disease onset. The mean relative afferent pupillary defect (RAPD) with filters was significantly greater in patients who developed rubeosis than in those who did not (0.9 vs 0.3 log units; p = 0.012). Using IRP, the pupillary diameters in the dark (maximum) and in the light (minimum) were significantly greater, the rate of pupillary constriction was significantly lower, and the latency of constriction was significantly greater in affected eyes than in unaffected eyes. The differences between affected and unaffected eyes in the IRP parameters of latency, rate, maximum, and minimum pupillary diameters were significantly greater in patients who developed rubeosis than in those who did not. Discriminant analysis of the IRP parameters correctly and statistically significantly identified rubeotic patients with 83% sensitivity and 95% specificity. An RAPD of > or = 0.6 log units was 83% sensitive and 70% specific in this regard. It is concluded that pupillary reactions are abnormal in many patients with acute CRVO, as measured by both pupillometric methods. The degree of these abnormalities has a relationship to the development of rubeosis, and might prove useful in planning the follow up of these patients or in deciding whether to apply panretinal photocoagulation. The neutral density filter test is readily available but subjective. IRP is more specific, objective, and suited to further development, but requires sophisticated equipment.