An important consideration in mathematical modeling and numerical simulation is that of validation of the models. This is particularly critical in thermal systems and processes because of the simplifications and idealizations usually needed to make the problem amenable to a solution, lack of accurate material property data, combined mechanisms, uncertainty in boundary conditions, and other complexities in the process. Modeling is needed for a basic understanding of the processes involved, as well as for providing accurate inputs for system design, control, and optimization. However, it is important to ensure that the numerical code performs satisfactorily for the chosen method and that the model is an accurate representation of the physical problem. These aspects are sometimes referred to as verification and validation, respectively, or simply as validation of the mathematical/numerical model employed. Unless the models are satisfactorily validated and the accuracy of the predictions established, the models cannot be used as basis for design and for choosing operating conditions to obtain desired system or product characteristics. Validation of the models is based on a consideration of the physical behavior of the results obtained, elimination of the effects of arbitrary parameters like grid and time step, comparisons with available analytical or numerical results for similar problems, and comparisons with experimental results on a physical model or a prototype. This paper considers some of the major considerations involved in validation of numerical simulation models for complex thermal problems. A few examples of practical problems are also presented. This topic is chosen because it was of particular interest to Professor Darrell Pepper and the author has had several interactions with him on complex thermal problems.