We analyze daily precipitation extremes produced by a six-member ensemble of the Pan-Arctic Weather Research and Forecasting (WRF) that simulated 19 years on the Coordinated Regional Climate Downscaling Experiment (CORDEX) Arctic domain for the Arctic summer. Attention focuses on four North American analysis regions defined using climatological records, regional weather patterns, and geographical/topographical features. We compare simulated extremes with those occurring at corresponding observing stations in the U.S. National Climate Data Center's Global Summary of the Day. Our analysis focuses on variations in features of the extremes such as magnitudes, spatial scales, and temporal regimes between regions. Using composites of extreme events, we also analyze the processes producing these extremes, comparing circulation, pressure, temperature, and humidity fields from the ERA-Interim reanalysis and the model output. Although the model's extreme precipitation is low compared to the observed one, the physical behavior in the reanalysis leading to observed extremes is simulated in the model. In particular, the reanalysis and the model both show the importance of moisture advection against topography for producing most of the extreme daily precipitation events in summer. In contrast, parts of Arctic western Canada also have a substantial contribution from convective precipitation, which is not seen in the other regions analyzed. The analysis establishes the physical credibility of the simulations for extreme behavior. It also highlights the utility of the model for extracting behaviors that are not easily discernible from the observations such as convective precipitation.