A close correlation in location, orientation and sense of slip is found between the Dobbs Ferry fault zone (DFFZ) and the 1985 Ardsley earthquake rupture (mb =4.0) in southern Westchester Co., N.Y. A 3 km portion of the DFFZ that had been previously recognized by Hall is now traced for 8 km from the Hudson River to Cameron’s Line. Detailed mapping reveals a northwest trending zone of intense fracturing and faulting associated with a set of aligned linear valleys. The fault zone is a tabular zone of discontinuous faults and fractures rather than a single through-going fault, and cumulative offset across it is very small, considering size and age of the structure. Fractures range from purely extensional joints, to small-displacement slickensided microfaults, to brittle faults with measurable offset, breccia, and gouge. Mesoscopic structural data argue for predominantly left-lateral motion on the DFFZ. Sinistral motion also predominated in the 1985 earthquake sequence. The accumulated displacement on the DFFZ shown by offset Paleozoic-age markers, however, is right-lateral (20–30m). Other northwest-striking brittle faults in the Manhattan Prong exhibit similar right-lateral offsets. This discrepancy along with other data suggest a multi-phase evolution, probably including a plate-boundary phase coinciding with Mesozoic rifting and a presently active intraplate phase. The more recent left-lateral phase probably involved fault formation from preexisting (Late Paleozoic-Mesozoic ?) joints. Preliminary results of this study have immediate application to earthquake hazard analysis and broad implications regarding intraplate seismogenesis. Firstly, faults with little or no accumulated displacement (“low-displacement faults”) can be the source of significant earthquakes. Thus, rates and recency of displacement on faults need to be reevaluated as criteria for earthquake potential. Secondly, the structural setting of intraplate seismogenic faults may provide a basis for segmentation, leading to limits on the maximum characteristic size of earthquakes, which could be much smaller than magnitude estimates derived from a single rupture of the entire fault.