While multiple in vitro methods have been developed, none are considered a complete replacement for the in vivo rabbit test for eye irritation classification. Instead, they are considered adequate for identifying seriously eye damaging chemicals and chemicals not requiring classification, based on GHS criteria for hazard classification and labelling. Due to the complexity of formulations, these mixtures are not included as reference chemicals in test method validation efforts. Herein we described two approaches to potentially reduce the use of animals for eye irritation testing of agrochemical formulations, both involving retrospective evaluations of existing data. First, we describe an evaluation of more than 160 agrochemical formulations to determine the usefulness and limitations of three in vitro ocular irritation test methods (Bovine Corneal Opacity and Permeability assay OECD 437, Isolated Chicken Eye Test method OECD 438, and EpiOcular eye irritation test OECD 492) to assign hazard classification for eye irritation potential. Second, we review the performance of the GHS concentration threshold approach for predicting eye irritation of agrochemical formulations based on the collective toxicities of individual formulation components. While there was insufficient overlap of formulations tested in multiple in vitro methods to evaluate a potential Defined Approach (DA), our results suggest that the individual in vitro methods and the concentration threshold approach were highly predictive of agrochemical formulations that would not require classification for eye irritation (negative predictivity ranged from 86% to 100%). Given the preponderance of agrochemical formulations that fall into this category, such an approach could significantly reduce the use of animals for this endpoint