Regional frameworks have been used extensively in recent years to aid in broad-scale management. Widely used landscape-based regional frameworks, such as hydrologic landscape regions (HLRs) and physiographic provinces, may provide predictive tools of hydrologic variability. However, hydrologic-based regional frameworks, created using only streamflow data, are also available and have been created at various scales; thus, relating frameworks that share a common purpose can be informative. In addition, identifying how the relative importance of variables change in governing streamflow with respect to scale can also be informative. The purpose of this study was to determine whether landscape-based frameworks could explain variation in streamflow classifications and in the hydrologic variables used in their creation. We also evaluated how climate and watershed-based variables govern the divergence of different flow classifications at two different scales. HLRs and physiographic provinces poorly predicted flow class affiliation within our study and for the entire USA, although physiographic provinces explained more variability. We also found that HLRs explained very little variation in individual hydrologic parameters. Using variables summarized at the watershed scale, we found that climate will play a larger role in influencing hydrology across the entire US, whereas soils may govern variation in hydrology at smaller scales. Our results suggest that predictor variables, developed at the watershed scale, may be the most appropriate at explaining hydrology and that the variables used in creating regional landscape-based frameworks may be more useful than the frameworks themselves. In addition, managers should be careful when using landscape-based regional classifications for stream management because the scale of their construction may be too broad to capture differences in flow variability. Copyright © 2011 John Wiley & Sons, Ltd.