A multi-phase program of model intercomparison is proposed to reduce uncertainties in the global carbon budget resulting from differences among atmospheric tracer transport simulations. The TransCom program is already well underway, having begun in 1993 and completed two ma- jor experiments. The results to date are summarized in a Technical Report and indicate that sig- nificant differences among the participating models need to be explored in much more detail if the intercomparison is to be useful in reducing uncertainties in the carbon budget of the atmo- sphere. A three-year program of further research is proposed. In the first year, a “calibration” ex- periment will be undertaken to allow quantitative evaluation of the realism of simulated tracer transport of SF6 by comparing the model results to observations of this anthropogenic trace gas. New model diagnostics will be incorporated into these runs which will permit a mechanistic analysis of transport processes that determine the outcome of the experiment. In the second and third years, sensitivity experiments will be designed and performed based on the calibration re- sults. These experiments will aid not only in reducing the uncertainties associated with differenc- es among the models, but also contribute to the understanding of processes that will aid future model development. A progress report will be delivered at a major international meeting in 1997, and a final report will be presented at another meeting in 1998.