PURPOSE: To evaluate the toxicity of a variety of the fluoroquinolone antibiotics on the ocular surface by using tissue culture models of corneal epithelial cells and conjunctival epithelial cells. METHODS: Immortalized conjunctival (CCC) and human corneal (HCE) epithelial cells were grown and when confluent the cells allowed to air dry for 1 hour. Medium was then replaced with 100 microL of one of the following: 1) Vigamox [moxifloxacin (0.5%: MX)]; (2) Zymar [gatifloxacin (0.3%: GA)]; 3) Quixin [levofloxacin (0.5%: LE)]; 4) Ocuflox [ofloxacin (0.3%: OF)]; 5) Ciloxan [ciprofloxacin (0.3%: CP)]; 6) medium (viable control); 7) "normal"/physiologic saline; 8) formalin (dead control). After one hour, 150 microL of MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazonium bromide was added and incubated for 4 hours. After decanting, precipitate was dissolved in 150 microL of isopropanol. Absorbance was determined at 572 nm. RESULTS: The lowest amount of cell death was associated with the viable control. All ophthalmic preparations showed both corneal and conjunctival cell toxicity. Aside from the viable control, normal saline showed the next lowest amount of toxicity. Of the topical ocular antibiotics tested, MX showed the least amount of toxicity. All of the other antibiotics tested were statistically indistinguishable from each other. CONCLUSIONS: All of the topical ocular antibiotics tested showed evidence of both corneal and conjunctival toxicity (MX < OF < or = LE < or = CP < or = GA), although only MX was statistically significant. Whether this finding reflects on in vivo wound healing remains to be determined. This model provides a rapid and cost-effective method to screen for surface toxicity of topical agents.