The disclosure relates to formulations and methods for the in vitro testing of ocular irritants. It was discovered that adding an antioxidant formulation to in vitro ocular irritation tests, including for example, a biochemical ocular irritation test, a reconstituted human corneal epithelium (RhCE) ocular irritation test and an excised eye depth of injury (DoI) test, substantially reduces the rate of false positives without diminishing test sensitivity, resulting in more accurately predicting ocular irritancy of test substances. More particularly, the disclosed method employs relatively high physiologic concentrations of one or more antioxidants that are normally present in tears. In a variation, much higher concentrations of one or more antioxidants may provide protection against in vivo exposure to ocular irritants.