During the past decade, paleoseismic studies done by many researchers in the coastal regions of the Pacific Northwest have shown that regional downdropping and subsequent tsunami inundation occurred in response to a major earthquake along the Cascadia subduction zone. This earthquake occurred almost certainly in 1700 A.D., and is believed by many to have been of M 8.5-9 or perhaps larger. In order to characterize the severity of ground motions from this earthquake, we report on a field search and analysis of seismically induced liquefaction features. The search was conducted chiefly along the banks of islands in the lowermost Columbia River of Oregon and Washington and in stream banks along smaller rivers throughout southwestern Washington. To a lesser extent, the investigation included rivers in central Oregon. Numerous small- to moderate-sized liquefaction features from the earthquake of 1700 A.D. were found in some regions, but there was a notable lack of liquefaction features in others. The regional distribution of liquefaction features is evaluated as a function of geologic and geotechnical factors in different field settings near the coast. Our use of widely different field settings, each in which we independently assess the strength of shaking and arrive at the same conclusion, enhances the credibility of our interpretations. Our regional inventory of liquefaction features and preliminary geotechnical analysis of liquefaction potential provide substantial evidence for only moderate levels of ground shaking in coastal Washington and Oregon during the subduction earthquake of 1700 A.D. Additionally, it appears that a similar conclusion can be reached for an earlier subduction earthquake that occurred within the past 1100 years, which also has been characterized by others as being M 8 or greater. On the basis of more limited data for older events collected in our regional study, it appears that seismic shaking has been no stronger throughout Holocene time. Our interpreted levels of shaking are considerably lower than current estimates in the technical literature that use theoretical and statistical models to predict ground motions of subduction earthquakes in the Cascadia region. Because of the influence of estimated ground motions from Cascadia subduction-zone earthquakes on seismic hazard evaluations, more paleoliquefaction and geotechnical field studies are needed to definitively bracket the strength of shaking. With further work, it should be possible to extend the record of seismic shaking through much of Holocene time in large portions of Washington and Oregon.