Within the last few years studies of relationships among faulting, fault creep, and seismicity in active areas such as California and Nevada have progressed rapidly, and have contributed greatly to our understanding of tectonics and the earthquake mechanism. This success prompted a field and literature search for similar phenomena in New York State and adjoining areas, a region of only moderate seismicity but of sufficient population density so that the seismic hazard should be thoroughly studied using all relevant information.Evidence for postglacial bedrock faulting, dated by offset striations, is common on outcrops of a belt of Paleozoic shales and slates that extends from near Hyde Park, New York, northerly along the east side of the Hudson and Champlain valleys, and continues northeasterly into Quebec at least as far as St. Georges. Faults tend to be in the plane of slaty cleavage with the south or east side upthrown in postglacial time. The observed faults are thus high-angle reverse faults with little or no postglacial strike–slip component. Displacement along a single fault is usually of the order of an inch or less, but commonly several such faults are seen in an outcrop only a few feet across. Thus, the faults seem very numerous, and cumulative displacement across a belt tens of miles in width could be quite substantial.An important question is whether this faulting is associated with glacial loading or unloading, with tectonic stresses, or with some other effect such as thermal changes, hydration, or a chemical process in the shales. Present data do not seem adequate to resolve this point, and a major purpose of this paper is to request similar information on other areas, particularly those adjoining the region of interest here. Casual observation of such faults would probably be dismissed by a geologist as slumping, frost heaving, or some other local disturbance in most cases if he were unaware of the regional pattern.Correlation between the pattern of observed faults and that of recent seismic activity is not highly convincing, but this could be because of inadequacies in either or both sets of data, so this question also remains open.