The acute dermal and ocular effects of chemicals are generally assessed by performing the Draize skin and eye tests, respectively. Because the animal data obtained in these tests are also used for the development and validation of alternative methods for skin and eye irritation, it is important to assess the inherent variability of the animal data, since this variability places an upper limit on the predictive performance that can be expected of any alternative model. The statistical method of bootstrap resampling was used to estimate the variability arising from the use of different animals and time-points, and the estimates of variability were used to determine the maximal extent to which Draize test tissue scores can be predicted.