This report reviews the geological basis for some recent estimates of earthquake hazards in the Cascadia region between southern British Columbia and northern California. The largest earthquakes to which the region is prone are in the range of magnitude 8-9. The source of these great earthquakes is the fault down which the oceanic Juan de Fuca Plate is being subducted or thrust beneath the North American Plate. Geologic evidence for their occurrence includes sedimentary deposits that have been observed in cores from deep-sea channels and fans. Earthquakes can initiate subaqueous slumps or slides that generate turbidity currents and which produce the sedimentary deposits known as turbidites. The hazard estimates reviewed in this report are derived mainly from deep-sea turbidites that have been interpreted as proxy records of great Cascadia earthquakes. The estimates were first published in 2008. Most of the evidence for them is contained in a monograph now in press. We have reviewed a small part of this evidence, chiefly from Cascadia Channel and its tributaries, all of which head offshore the Pacific coast of Washington State. According to the recent estimates, the Cascadia plate boundary ruptured along its full length in 19 or 20 earthquakes of magnitude 9 in the past 10,000 years; its northern third broke during these giant earthquakes only, and southern segments produced at least 20 additional, lesser earthquakes of Holocene age. The turbidite case for full-length ruptures depends on stratigraphic evidence for simultaneous shaking at the heads of multiple submarine canyons. The simultaneity has been inferred primarily from turbidite counts above a stratigraphic datum, sandy beds likened to strong-motion records, and radiocarbon ages adjusted for turbidity-current erosion. In alternatives proposed here, this turbidite evidence for simultaneous shaking is less sensitive to earthquake size and frequency than previously thought. Turbidites far below a channel confluence, instead of representing the merged flows from two tributaries, monitor the dominant tributary only. Sandy beds low in the turbidites, instead of matching from channel to channel, permit divergent stratigraphic correlations; and rather than approximating strong-motion seismograms, the sandy beds more likely record processes internal to the generation and transformation of subaqueous mass movements. The age adjustments, instead of supporting other evidence that all the northern ruptures were long, are uncertain enough to accord with variation in rupture mode, and this variation improves agreement with onshore paleoseismology. Many of the turbidites counted as evidence for frequent earthquakes on the southern Cascadia plate boundary may instead reflect nearness to steep slopes. This report is meant to aid in the updating of national maps of seismic hazards in Canada and the United States. It offers three main conclusions for consideration at a U.S. hazard-map workshop slated for March 21-22, 2012: If giant earthquakes are the norm for the plate boundary offshore southern Washington, the strongest paleoseismic evidence for this rupture mode is the average earthquake-recurrence interval of about 500 years that is evidenced both offshore in lower Cascadia Channel and onshore at estuaries of southern Washington and northernmost Oregon. The plate boundary offshore southern British Columbia and northern Washington may be capable of producing great earthquakes at an average interval as short as 300 years that is evidenced mainly onshore. Review of more of the turbidite evidence now in press may clarify implications for the hazard maps. Further work on the deep-sea turbidites could target sedimentary processes and chronological uncertainties that may affect the turbidites' sensitivity to fault-rupture lengths and recurrence rates.