We use output from a suite of Regional Climate Models (RCMs) applied under the North American Regional Climate Change Assessment Program (NARCCAP) to quantify possible changes in metrics of climate impacts in the Midwest for the middle 21st century (2041–2062). RCM simulations of the historical period indicate a large positive bias in growing season length, but generally good agreement with observationally derived estimates of metrics such as the mean summertime maximum and apparent temperatures and number of cooling degree days. There is a tendency towards intensification of thermal extremes by 2041–2062 that is equal to or exceeds the bias in the ensemble mean during the historical period (1979–2000). For example, the number of days in the Chicago region each year with temperatures in excess of 32.2 C (90 F) is doubled by the mid-century. The ensemble average estimates for 1979–2000 from the RCMs is negatively biased in terms of annual, spring and summer precipitation (the wettest pentad) and mean dry day duration, but is positively biased in terms of total precipitation accumulated on the 10 wettest days of the year. The RCM simulations indicate an increased likelihood of extreme hydroclimate events that again are equal to or exceed the historical biases. For example, the wettest pentad and total accumulation on the top 10 wettest days of the year are projected to increase by~ 10%. Preliminary results for icing and extreme wind events indicate no change in the magnitude of extreme wind events, and perhaps slight reductions in icing risk.