A program of shallow drilling, trenching, outcrop mapping, and seismic-reflection acquisition in southern Illinois, just north of the New Madrid seismic zone, has produced the most comprehensive documentation to date of Quaternary-age faulting in the northern Mississippi embayment. Detailed structural cross-sections over five northeast-trending faults that continue from the Fluorspar Area fault complex southward into the embayment (Massac County, Illinois) indicate narrow grabens into which latest Cretaceous, Paleocene, and Eocene sediments were dropped and protected from erosion. The bounding faults were active during Neogene through middle Pleistocene time; in the deepest graben, the Late Miocene-Pleistocene-age Mounds Gravel is downthrown 150 m. Notably, definitive faulting of Wisconsinan loess or Holocene alluvium is not observed at any site, which would indicate that the faults have been inactive for at least 55 ka (basal loess ages) to 128 ka (youngest Illinoian age). Seismic profiles reveal that faults in Quaternary sediments penetrate Paleozoic bedrock, thus indicating that these are tectonic faults and not products of landsliding, solution collapse, or other merely near-surface processes. The faults exhibit vertical to steeply dipping normal and occasional reverse displacements that outline a variety of structures, including a series of narrow grabens. Tilting or folding of strata associated with faulting is also locally observed. At one site, observed faulting is superimposed over a prominent bedrock anticline, which may itself be a result of the Quaternary deformation that created the steep normal and reverse faults. Our results indicate that shallow sediment and bedrock faulting extends beyond the localized fault zones, implying that Quaternary deformation of the northern Mississippi embayment is more pervasive than formerly believed. The trend of Quaternary-age structures in southern Illinois is in line with the northeast-trending New Madrid seismic zone and is parallel to other geologically active features in the region, such as the Commerce geophysical lineament; the lack of observed Holocene movement, however, suggests that the faults in our study area are not currently active and thus may not contribute to seismic hazard. We propose a dynamic structural model that suggests a mechanism by which seismicity and active (Holocene) faulting have shifted within the central Mississippi Valley (away from the Fluorspar Area fault complex) over the last several 10,000's of years.