For regional climate change applications, it is often necessary to utilize downscaling tools, which rely on accurate simulation of synoptic and/or continental scale climate phenomena. In most studies, accurate simulation of large-scale features is assumed, although such assumptions may be invalid. We evaluate these assumptions by examining the output from two commonly used coupled oceanic-atmospheric GCMs: the Hadley Center's climate model (HadCM3) and the Canadian Climate Center climate model (CGCM2). The evaluation is performed for the Midwestern region of the USA and consists of three steps. First, the ability of the models to simulate the daily North Atlantic Oscillation (NAO) and Pacific North American pattern (PNA) is evaluated. Second, the Kirchhofer map-type classification method is used to examine GCM reproduction of the synoptic scale climate in terms of means and variances of seasonal map type frequencies and map-type coherence, persistence, and progression. Lastly, the links between the phases of the NAO and PNA and the map-types are examined. In each case, the models are evaluated over a recent period (1990-2001) relative to the NCEP/NCAR reanalysis. This approach allows direct comparison of observations and simulations. To provide a context for the differences between observations and simulations, we also examine an earlier observed period (1953-1964) and a future modeled period (2030-2041). Preliminary results for HadCM3 indicate during the reference period (1990-2001), modeled map-types are similar to those in the NCEP/NCAR data in terms of frequency, coherence, persistence, and progression, although the most common map-type occurs more often in the GCM simulation, suggesting that HadCM3 may underestimate the total variability in the synoptic-scale circulation. However, despite the relatively high degree of correspondence between the observed and simulated teleconnection indices and synoptic types in the study area, differences in the period 1990-2001 were greater than either (1) differences in reanalysis data between 1953-1964 and 1990-2001 or (2) differences in HadCM3 between 1990-2001 and 2030-2041.