From December 1811 to February 1812, four large earthquakes (m_b≥7.0) occurred in the New Madrid (Missouri) Seismic Zone (NMSZ). These have been the largest historical earthquakes in eastern North America. Although this area has been the focus of considerable seismological research, estimates of the repeat time of large-magnitude seismic events remain poorly constrained. Past estimates were primarily based on earthquake-frequency statistics and on paleoseismology studies. Johnston and Nava (1985) compiled historical seismicity data covering about 180 years and instrumental data covering 10 years, and they concluded that the repeat time for large-magnitude events (m_b≥7.0) is between 500 and 1,100 yrs. However, this estimate is based on assumptions that the data set is representative of the seismicity of the region over the past 1,000 yrs, and that the relation between earthquake frequency and magnitude is constant (Johnston and Nava, 1985). Because these assumptions cannot be verified, this estimated recurrence interval of 550-1,100 hrs must be considered tentative (S. G. Wesnousky and L. M. Leffler, written commun., 1991). Investigation of exploratory trenches across the Reelfoot scarp in northwestern Tennessee documented the only unequivocal Holocene surface faulting in the upper Mississippi embayment (Russ and others, 1978; Russ, 1979). Fluvial sediment younger than about 2,250 yrs old is faulted, and the net vertical displacement is more than 3 m. Stratigraphic relations indicate at least two episodes of faulting occurred between about 2,250 yr B.P. and the 1811-12 events to estimate an average recurrence interval of less than 600 yrs for large-magnitude earthquakes in the NMSZ. However, inasmuch as Russ (1970) found no evidence for any historical offset on the Reelfoot scarp, the relation between the Reelfoot scarp and large paleoeartchquakes in the NMSZ has not been clearly established. The development of widespread liquefaction features suring the 1811-12 earthquake series (Obermeier, 1989, Obermeier and others, 1990) and the probably development of similar features during previous large-magnitude seismic events (m_b≥6.2, Nuttli, 1982) have been the basis for several attempts to document the history of paleoliquefaction. Haller and Crone (1986) found evidence of only one episode of sand-blow development in exploratory trenches in late Pleistocene alluvium in eastern Arkansas and concluded that this liquefaction event was probably associated with the 1811-12 earthquake series. Saucier (1989) reported evidence of three liquefaction events in the past approximately 1,000 yrs in an exploratory trench in eastern Arkansas. On the basis of the apparent absence of post-depositional erosion separating the three sand-blow deposits, Saucier (1989) concluded that they formed in a relatively short period of time, probably during the 1811-12 earthquake series. Similarly, Schweig and Marple (1991) found evidence of only recent (probably 1811-12) liquefaction in exploratory trenched on late Wisconsin braided-stream deposits in southeastern Missouri. Leffler and Wesnousky (1991) and Wesnousky and Leffler (written commun., 1991) examined tens of kilometers of recently excavated drainage ditches in late Wisconsin braided-stream deposits in eastern Arkansas and found no evidence for prehistorical liquefaction events during the last 10,000 yrs. In contrast, Saucier (1991) estimated an average recurrence interval about 470 yrs on the basis of historical ages of liquefaction in the NMSZ. The lack of similar evidence from other sites implies a prehistorical liquefaction-producing seismic event in southeastern Missouri that was considerably smaller than the 1811-12 earthquake series. The apparent absence of paleoliquefaction features in late Wisconsin fluvial deposits in the zone of most intense 1811-12 liquefaction indicates that the repeat time of large (m_b≥7.0) seismic events in the NMSZ is at least 10,000 yrs (Saucier, 1991; Wesnousky and Leffler, written commun., 1991). The present study was undertaken to verify this conclusion by documenting the record of liquefaction in late Wisconsin fluvial deposits along the Obion River, in northwestern Tennessee, This region was identified by Obermeier (1989) as an area deserving further study because of its proximity to the NMSZ and to areas of historical liquefaction.