Earthquake hazard analyses in Italy are mainly based on historical seismic catalogs. However, the completeness of the Italian catalog, which spans more than 2000 years, progressively decreases going back in time. In the Fucino basin, historical research has not produced evidence of large earthquakes before the January 13, 1915, event (M•. = 7.0; 33,000 people killed). This event generated surface faulting along at least two parallel Quaternary normal faults bounding the eastern side of the basin (the Celano-Gioia and Parasano-Cerchio faults), over an end-to-end length of about 23 km, with a downthrow of 30-90 cm toward southwest. To improve geological aspects of the seismic history of this area, we trenched one of the coseismic scarps along the Celano-Gioia fault. At a trench site, near the village of San Benedetto de' Marsi, the fault traverses the area formerly covered by the Fucino Lake, drained for farming in 1875 A.D. The evidence collected in the trenches suggests at least two large events in historical times before 1915. Radiocarbon dates constrain the age of event B between the 6th and the 9th centuries A.D., possibly coinciding with the event felt in Rome in 801 A.D. [Molin and Guidoboni, 1989]. Event A probably occurred after the 10th century A.D. and almost certainly before the 1349 A.D. earthquake, because historical studies show (1) that event A was not produced during the earthquake sequence that occurred in central Italy in 1349 A.D. [Molin and Guidoboni, 1989] and (2) that the post-1349 seismic catalog can be regarded as complete for surface faulting earthquakes. Surface displacements determined for events A and B, compared to the 1915 surface faulting, indicate that M > 6.5 earthquakes were associated with each rupture. These observations suggest that the return period for large earthquakes in the Fucino basin is shorter than that indicated by the historical record alone. The Fucino example clearly shows that the exceptionally long historical record in Italy is still inadequate for a comprehensive seismic hazard characterization. Paper number 95JB02852. 0148.0227/96/95JB_02852505.00 33,000 people killed, M.• = 7.0, rn•, = 6.8 [Margottini and Screpanti, 1988]). Because the Italian catalog is commonly considered complete for large (M > 6.5) earthquakes in the last !000 years, this suggested to previous authors [e.g., Serra, ! 987; Ward and Valensise, 1989] that events like that of 1915 should have a return period greater than 1000 years. However, Cartara et al. [1989] point out the poor historical record for the Fucino region during the Middle Ages (6th-15th century A.D.). In addition, Giraudi [1988b] recognized a historical surface faulting event that probably occurred during the 5th to 14th centuries A.D. To better define the seismic history of the area, we chose a site for trench investigations along one of the Holocene scarps rejuvenated during the 1915 event [Serva et al., 1988; BIumetti et al., 1988]. At this site, near the village of San Benedetto de' Marsi (Figure 2), the coseismic scarp is...