In a previous study of the extratropical Northern Hemisphere wintertime circulation, significant variations were described in the occurrence of anomalies that persist beyond the durations associated with synoptic-scale variability (“persistent anomalies”). The present study extends that work by identifying the typical structures of persistent anomalies, focusing on whether persistent occur in certain key regions (the central North Pacific (PAC), the eastern North Atlantic (ATL) and the northern Soviet Union (NSU)] are related to recurrent large-scale flow patterns. For each region, detailed comparisons are provided of the flow patterns associated with persistent positive and negative anomaly cases, and with the patterns obtained in other observations and theoretical studies of persistent phenomena. The results provide evidence for the recurrence of certain preferred anomaly patterns. To a first approximation, the composite anomaly patterns of the positive and negative cases for a region can be described as opposite phase of the same basic pattern. This pattern is identified as. the primary regional pattern of low-frequency variability. For the PAC and ATL means the primary patterns resemble, respectively, the Pacific-North American (PNA) and Eastern Atlantic (EA) teleconnection patterns described by Wallace and Gutzler. The majority of the persistent anomaly cases in each region are related to episodes of unusually large amplification of the of the primary patterns, with one phase frequently associated with blocking and the other with an abnormally strong zonal flow. The persistent flow anomalies are also typically accompanied by strong tropospheric temperature anomalies having patterns mainly in-phase with the height anomaly patterns, significant changes in the locations and intensities of the major surface centers of action and pronounced shifts in the locations of maximum storm activity. Most of the temporal variability of the primary patterns is contributed by low-frequency, intraseasonal fluctuations (integral time scales of a few weeks). The interannual variance in both the ATL and NSU patterns is consistent with the level expected from the sampling of these relatively short-term fluctuations; however, a significant fraction of the interannual variance of the PAC pattern appears to be above the level expected from sampling variability.