This paper argues that the geologic record, and the late Quaternary history in particular, is a far more valuable tool in estimating seismicity and associated seismic hazard than has generally been appreciated. Those parts of the world with the longest historic records of earthquakes — some 2,000 yr for Japan and the Middle East and 3,000 yr for China — are the areas that should give us the greatest pause in using historic records for extrapolations, because earthquakes in these regions show surprisingly large long-term temporal and spatial variations. The very short historic record in North America should, therefore, be used with extreme caution in estimating possible future seismic activity. The geologic history of late Quaternary faulting is the most promising source of statistics on frequencies and locations of large shocks.
Seismotectonic relationships in California, where the one-to-one correlation between large earthquakes and active faults is well documented, also apply to other parts of the world to a greater extent than has generally been recognized. The same is true for the frequent evidence of surface faulting associated with both large and small shallow earthquakes. The long history of Turkish earthquakes illustrates marked temporal changes in spatial distribution of seismicity, but all major seismic areas in Turkey could easily have been identified even in the absence of historic records by field studies of Quaternary faulting. Central Japan has so many Quaternary faults that seismic hazard must be considered relatively uniform and widespread. It is disconcerting that some of the largest Japanese earthquakes have occurred on seemingly innocuous faults; conversely, some of the largest and most spectacular faults, such as the Median Tectonic Line and the Itoigawa-Shizuoka Tectonic Line, have caused no major earthquakes within the historic record. Abundant evidence of Quaternary and probable Holocene displacements on these two faults, however, suggests that they are likely sources for future large events. China likewise shows close association between active faults and major earthquakes; perhaps in no other part of the world are Quaternary faults more spectacular. A segment of the Philippine fault that was the locus of a recent large earthquake demonstrates that major active faults can be identified adequately in the field even in areas of tropical vegetation. Throughout the world, thrust faults present a special problem in seismic hazard evaluation, because their configurations and degrees of activity are more difficult to determine in the field than those of strike-slip and normal faults. Particularly for thrust faults, trenching and bore-hole techniques are essential exploration tools.
The most important contribution to the understanding of long-term seismicity, which is critical to the siting and design of safe structures and to the establishment of realistic building codes, is to learn more — region by region — of the late Quaternary history of deformation.