There are currently available many microcomputer-based hazard response models for calculating concentrations of hazardous chemicals in the atmosphere. The uncertainties associated with these models are not well-known and they have not been adequately evaluated and compared using statistical procedures where confidence limits are determined. The U.S. air force has a need for an objective method for evaluating these models, and the proposed research will provide a framework for performing these analyses and estimating the model uncertainties. As part of this research, available models and data sets will be collected, methods for estimating uncertainties due to data input errors and stochastic effects will be developed, a framework for model evaluation will be put together, and preliminary applications using test data sets will take place.