In a randomised study 25 patients over 50 years of age with non-ischaemic central retinal vein occlusion (CRVO) were assigned to either a treatment (14 patients, isovolaemic haemodilution) or a control group (11 patients, no treatment). After three months eight eyes in patients with haemodilution improved, whereas none of the untreated eyes had better visual acuity (p less than 0.01). Thirteen haemodiluted and 11 control patients could be observed for one year. Six eyes of the haemodiluted patients retained a better visual acuity, whereas no improvements had occurred in the control group (p less than 0.025). In fluorescein angiography the lowering of the packed cell volume to 35-32% accelerated the time of maximal venous filling (tmvf) from 17.4 (SEM 1.4) s to 11.4 (SEM 0.9) s (p less than 0.005). In patients with non-ischaemic CRVO without treatment the passage time did not change. The shortened tmvf reflects a higher blood velocity. Thus isovolaemic haemodilution improves the visual prognosis in non-ischaemic CRVO probably by inducing a higher blood fluidity, which results in higher blood velocity, at least in areas of compromised retinal microcirculation.