Many trace atmospheric constituents affect the radiative budget of the atmosphere. The Kyoto protocol includes carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), perfluorocarbons (PFCs), hydrofluorocarbons (HFCs), and sulfur hexafluoride (SF6). A variety of studies suggest that potential carbon sinks and opportunities to limit emissions of other greenhouse gases (GHGs) may reduce the cost of control. They also point to the risks of failing to control gases with very long lifetimes. Few studies have yet considered an integrated evaluation of the costs of multi-gas control strategies, or the implications of reductions in different mixes of GHGs on atmospheric composition, climate, and ecosystems. Using the MIT Integrated Global System Model (IGSM) as recently modified we examine multi-gas control as envisioned by the Kyoto protocol, exploring the costs of emissions reduction and the consequences for the atmosphere, climate, and ecosystems. The basic components of the IGSM are an Emissions Prediction and Policy Analysis (EPPA) model, a Natural Emissions Model, a coupled Atmospheric Chemistry and Climate Model, and a Terrestrial Ecosystems Model. We stop short of quantifying damages in monetary terms. Our ecosystem results illustrate tradeoffs that result from different control strategies. We find that inclusion of sinks and abatement opportunities for gases other than CO2 could reduce the cost of meeting the Kyoto agreement by 60 percent. Assuming the protocol is extended unchanged to 2100, we find little difference in climate and ecosystem effects between 2010 and 2100 for a strategy that achieves the required reduction with a multigas as compared to a CO2-only strategy. Under a more aggressive policy, increasing the reductions in Annex B countries and extending reductions to the rest of the world after 2010, significant differences in effects develop between the two strategies. This latter result indicates that 100-year GWPs as currently estimated fail to capture important time horizon and climate-chemistry interactive effects, and this failure can be significant for policy.