Every year, numerous firefighter burn injuries and deaths occur in the U.S. Most of these burn injuries and deaths occur due to inadequate performance of firefighters' thermal protective clothing under various thermal exposures including hot water and steam exposures. Therefore, it is necessary to develop models for conveniently predicting the hot water and steam protective performance of fabrics used in firefighters' clothing. This study aims at developing a new approach for creating models to predict the hot water and steam protective performance of fabrics used in firefighters' clothing. This aim was achieved by fulfilling two objectives – firstly, by characterizing the performance of fabrics; secondly, by empirically modeling the performance of fabrics. To accomplish these objectives, physical properties (eg, thickness, air permeability) and the performances of single‐ and multilayered fabrics used in firefighters' clothing were measured. The measured data were statistically analyzed to identify the key fabric properties affecting the performance. Using these key properties, multiple linear regression (MLR) and artificial neural network (ANN) models were developed. It has been found that thickness, air permeability, and evaporative resistance are the key properties to affect the performance, and ANN is the best‐fit model to predict the performance. The approach suggested in this study could be used to develop state‐of‐the‐art models for predicting the performance of wider range of fabrics under hot water and steam exposures. These models could inform firefighters about the potential protective performance of their clothing before working in hot water and steam exposures.