Regional classification techniques can help in identifying spatial structure in river flow variation. As such, a framework can be provided around which to test hypotheses, making regional classification an important tool for hydrological research, particularly over large and complex geographical domains. Here, the results of a regional classification of monthly river flow for the northern North Atlantic region over the period 1968–1997 are presented. Importantly, regional classes are identified on the basis of inter-annual variation in river flow time series, rather than the more commonly used intra-annual flow regime. This is beneficial because it provides an improved basis for the investigation of regional-scale inter-annual trends in river flow: an important aim given the potential for future changes in river discharge linked to climate change and land-use change. Seven regions (four in northern Europe and three in eastern North America) are identified using hierarchical cluster analysis (Wards algorithm). Analysis of the hydrological homogeneity of the emergent river flow regions reveals them to be useful groupings to frame further analyses. Investigation of the statistical characteristics of each mean regional time series follows, including the annual flow regime, time series variability, and the presence of trends (as indicated by the Mann-Kendall test). Key findings identified by the latter analysis include an increasing trend in late winter discharge for some northern European regions for 1968–1997, but overall few statistically significant trends. Copyright © 2010 John Wiley & Sons, Ltd.