The purpose of the present study was to develop an optimal freezing method for cryopreservation of human donor corneas for transplantation. Three groups of ten human donor corneas each were cryopreserved using cooling rates of 0.5 degree C/ min, 1 degree C/min and 9 degrees C/min. The freezing medium contained 10% fetal calf serum and 7% dimethylsulphoxide. Ten additional human donor corneas were used as controls. Endothelial cell survival after complete thawing was assessed by calculating the mean endothelial cell density and percentage of non-viable endothelial cells from vital staining and scanning electron micrographs. Significant differences in endothelial cell survival between all groups were detected by analysis of variance (p < 0.001), but paired contrast found no real differences between corneas frozen at 0.5 degree C/min or 1 degree C/min. Nevertheless, 60% of the corneas cryopreserved at 1 degree C/min but only 10% of those cryopreserved at 0.5 degree C/min fulfilled both requirements to be considered suitable for transplantation (endothelial cell density > 2000 cells/mm2 and less than 10% dead cells). None of the corneas frozen at 9 degrees C/min fulfilled these requirements. Mean endothelial cell density of corneas frozen at 1 degree C/min was 2084 cells/mm2 (range 2020 to 2630 cells/mm2). Except for isolated corneas frozen at 0.5 degree C/min, only corneas cryopreserved at a cooling rate of 1 degree C/min achieved satisfactory endothelial cell survival for their use in transplantation. However the significant interindividual variability among corneas frozen at 1 degree C/min (ANOVA: p < 0.001) prevents prediction of the condition of the cornea after the complete cryopreservation process. Until this high variability can be drastically reduced, systematic corneal cryopreservation in eye banking remains unattainable.