Forecast skill of the synoptic-scale flow over the Arctic is evaluated by calculating a standardized anomaly of area-averaged root mean square error (RMSE) of day-5 forecasts of 500-hPa geopotential height over the Arctic during summer (June–August) for the 2007–2017 period in the 11-member NOAA GEFS reforecast dataset v2. Periods characterized by exceptionally high values of standardized anomaly of area-averaged RMSE are referred to as low-skill periods. ACs occurring during low-skill periods are identified using an ERA-Interim AC climatology. Forecast skill of ACs occurring during low-skill periods is evaluated by calculating the RMSE of the intensity of the ACs for forecasts in the GEFS reforecast dataset v2. The forecasts are of 120-h lead time and valid at the time of the lowest sea level pressure of the ACs when located in the Arctic during low-skill periods. ACs characterized by relatively high values of RMSE are referred to as low-skill ACs. Composite analyses are performed using the ERA5 on intense low-skill ACs in order to examine various features and processes governing the evolution of intense low-skill ACs. The composite analyses indicate that intense low-skill ACs intensify downstream of a mid-to-upper-tropospheric vortex in a region of relatively strong lower-to-midtropospheric baroclinicity, lower-to-midtropospheric ascent, tropospheric-integrated vapor transport, and upper-tropospheric divergence. The composite analyses suggest that a combination of baroclinic processes and latent heating play important roles in the intensification of intense low-skill ACs.