Climate change assessments often utilize only a small number of climate stations within the assessment region. Consequently, the selection of representative stations that capture the spatial variability of climate is an essential first step of the assessment. Furthermore, climate observations and scenarios frequently need to be linked with other (e.g. agricultural, biological, socioeconomic) datasets available at varying spatial resolutions in order to assess regional-scale impacts and vulnerability. This study illustrates the utility of well-known objective climate classification techniques in facilitating regional-scale climate change assessments, using the Upper Great Lakes region (UGLR) of the United States as an example. A combination of principal components analysis and cluster analysis was employed to group climate stations from the US Historical Climatology Network into a small number of climate regions. The regionalization initially was performed on long-term averages of monthly maximum and minimum temperatures and precipitation for 1971–2000, and then applied to two earlier time periods (1911–1940 and 1941–1970) to assess the sensitivity of the regionalization to the development period. The spatial variability of average monthly temperature and precipitation in the UGLR was found to be effectively represented by seven climate types. Potential applications of the regionalization for climate change assessments are illustrated. First, Euclidean distance, employed within a geographic information system (ArcGIS), provided a convenient means of assigning political units (counties) within the UGLR to the climate regions to facilitate the integration of point-scale climate information with spatially-aggregated datasets. Second, a minimum number of climate stations for an assessment were selected based on the climate regionalization. Finally, the regionalization was used to evaluate the spatial representativeness of a subset of climate stations that were initially chosen based on the completeness of daily observations.