Previously, we have simulated the atmospheric transport and fate of mercury emissions in North America and derived estimates of ambient concentrations and dry and wet deposition of mercury. In this study we quantify sensitivity of the derived estimates to model input parameters that we believe have the largest potential to influence model estimates. We vary five input parameters: emission speciation, Hg(II) dry deposition velocity, precipitation amount, concentration of redox species, and Hg(II) boundary conditions, within their plausible range of values. Our results show that emission speciation has the largest influence and Hg(II) boundary conditions have the smallest influence on the derived estimates. The sensitivity of simulated wet deposition to emission speciation and redox species concentration is non-linear and varies by region. In regions with low wet deposition (5--15 μg m-2 yr-1), emission speciation and chemistry show comparable influence, whereas in regions with high wet deposition (15--30 μg m-2 yr-1), emission speciation shows greater influence than chemistry. The interregional differences in sensitivity suggest that different pathways control total wet deposition for different regions. While in our previous study we evaluated the modeling system against observations, the sensitivity studies described in this paper enabled us to obtain new insights on atmospheric mercury by focusing on the dynamics of the system, i.e., response of the system to variation in its inputs. This analysis is essential before model-simulated results are used to investigate source-receptor relationships. Our findings also indicate that there is a critical need to get additional data on mercury speciation of major emission sources. ¿ 1999 American Geophysical Union