Historical simulations from 15 models participating in Phase 6 of the Coupled Model Intercomparison Project (CMIP6) are evaluated for their ability to reproduce several important characteristics related to short-duration dry spells over the northeastern United States and their accompanying large-scale meteorology. A dry spell is defined as consecutive days when the area-averaged precipitation is <2mm. The models are compared to observations in terms of capturing the correct number of dry spells, their representation of the associated large-scale meteorological patterns (LSMPs), and their ability to represent upper-level zonal flow, as well as low-level moisture. The LSMPs are identified using k-means clustering on the 500-hPa stream function anomaly data during the dry spell events for each data product. There are substantial differences from observations in terms of the models’ ability to capture the correct number of dry spell events, with models tending to overestimate the number of short events (1 or 2 days long) and underestimate the number of long events (12 days or longer). Furthermore, while the LSMPs determined in the models are visually similar to those from observations, the location and magnitude of specific troughs and ridges within those patterns, as well as the seasonality of those patterns, can differ substantially from corresponding observations. Several metrics and a model weighting scheme are devised based on regional characteristics of interest in order to facilitate comparisons between the models and to rank the models based on performance in several key categories. Ongoing work investigates how model climatological mean biases in the 500-hPa stream function and 250-hPa zonal wind fields, as well as jet stream positioning, influence the model biases in the LSMPs. This work is of relevance and importance for facilitating model improvement and future model development.