Monte Carlo uncertainty approaches are being increasingly used in air quality modeling in order to estimate the magnitudes of the uncertainties in model outputs due to uncertainties in model inputs. The results can be further analyzed to determine the input variables that have the greatest effect on the output uncertainties. It is first necessary to clearly define the model outputs that will be studied, and then define the probability density functions (usually log-normal or normal) and the mean and variances of the uncertainties in model inputs, parameters, and alternate algorithms. Strong correlations (magnitudes greater than 0.6 or 0.7) between input variables should be accounted for. Then the modeling system is run 100 or more times by randomly sampling from all input variables for each run. An overview of the various problems and issues that arise, as well as some examples of results of Monte Carlo uncertainty analysis, are given in this paper. The primary example that is discussed concerns the BEIS-3 biogenic emissions model coupled with three photochemical grid models (MAQSIP, UAM-V, and URM), as applied to the OTAG geographic domain in the Eastern U.S. for three multi-day episodes in 1995. It is shown that the uncertainties in temperature are the primary influence on uncertainties in BEIS-3 emissions estimates.