A diagnostic analysis examined precipitation and surface air temperature in the central United States for transient simulations from the Coupled-Model Intercomparison Project (CMIP)2+ database. We compared seasonal averages based on the last 30 years of the control simulations with observed values for the present-day climate. For precipitation, comparisons were made for certain features of the circulation that are correlated with observed precipitation. These include the Great Plains low-level jet and the upper tropospheric jet stream; high correlations between monthly precipitation anomalies and circulation anomalies in these regions are found in the observational record. We examined both whether the models accurately reproduced the mean flow patterns and also whether they reproduced the high correlations between precipitation and circulation anomalies. Preliminary results of our analysis of the control simulations suggest the following. One, most models provide a rather accurate simulation of precipitation for winter, spring, and summer. Two, all models but one have a substantial negative bias for fall precipitation. Three, the mean flow patterns in the low level and upper tropospheric jet regions are simulated with a moderate degree of accuracy. Four, there is considerable scatter in the correlations between precipitation and circulation anomalies; for a number of models the correlations differ substantially from observed values. Fifth, seasonal temperature differences between models and observations are generally +/- 2°C except for a few models in winter and spring where differences as large as 4°C were found. Differences between the control simulation and the transient simulation around the time of CO2 doubling (years 65-75) were examined. The precipitation changes are quite varied and include both increases and decreases. Model results are most varied in the summer, with changes ranging from a nearly 20% decrease to a 15% increase. Interestingly, none of the precipitation changes in the transient simulations are greater than the envelope of natural variations observed in the control simulation. For temperature, the models are also quite varied with warming varying by about a factor of 2-3 among the models.