The effect of artificial tear solutions on the corneal epithelial barrier was evaluated by measuring corneal uptake of 5,6 carboxyfluorescein (CF) after exposure of rabbit corneas to various formulations in a conjunctival cup. Four tear solutions containing 0.01% benzalkonium chloride (BAC), a formulation containing 0.001% Polyquad, a contact lens re-wetting solution containing 0.004% thimerosal, and each of the above preservatives in balanced salt solution (BSS) were evaluated. Four non-preserved solutions were also tested. After treatment with the solutions corneas were exposed to the CF, removed from the eyes and dialyzed in balanced salt solution. The CF concentration in the dialysate was measured by fluorometry. Solutions containing 0.01% BAC caused a 9.24 to 99.28 fold increase in CF uptake as compared to control. Solutions preserved with Polyquad or thimerosal caused only a 0 to 4 fold increase in CF uptake while non-preserved solutions caused no change compared to control. Corneas prepared for transmission electron microscopy using fixative containing ruthenium red exhibited damage which correlated well with CF uptake; the ruthenium red penetrated the epithelium to the basal cell layer after corneal exposure to solutions containing BAC while only superficial cell layers were stained after exposure to the other test solutions. The method used in this study allows statistical comparison of artificial tear formulations. The data show that patients with severe dry-eye who use artificial tears frequently should avoid tear solutions containing BAC and that non-preserved solutions are preferable for treatment of these patients.