This paper explores the reliability of intermedia transfer models as a function of both performance (i.e., validity) and specific decision objectives. The performance of selected models is gaged by measuring the error associated with practical model applications. Published volatile organic contaminant leaching and volatilization data are compared with deterministic and stochastic modeling results. The observed intermedia transfer modeling error ranges from 6 to 894%. A hypothetical residential exposure scenario is developed for inhalation of benzene vapors volatilized from soil. Experimentally determined and modeled vapor flux rates are translated into time-averaged air concentrations using a simple box model for air mixing. Health risk estimates based on experimental data are compared with the model-based estimates. Contributions to the overall uncertainty in predicted risk from exposure parameter uncertainty is compared with contributions from volatilization model parameter uncertainty. Modeling parameter uncertainty contribution is found to be dominated by the parameter uncertainty contributed from the exposure evaluation. Implications of error and uncertainty for intermedia transfer model usability in human exposure assessment are explored by considering 1) the sensitivity of predicted health risk values to model performance and 2) the consequences of making an incorrect decision.