The Niscemi landslide occurred on 12 october 1997, without any apparent connection to a significant destabilising event, at the end of a long dry season and two to three days after intense, but not exceptional, rain storms (71 mm on October 9, and 26 mm on October 10, respectively). It was a landslide of slow and moderate displacement (having an estimated maximum velocity of 1 m/h and displacement up to 2-4 m). It involved a highly asymmetric unstable mass and two orthogonal slopes in a sole movement, covering a surface of about 2 km <sup>2</sup> , with a main scarp of about 2,5 km and with ill-defined lateral borders. The unity of the movement is testified by the little deformation existing at first along all the main scarp during the pre-paroxysmal stage. The sliding surface developed almost entirely within a homogeneous clayey formation (Upper Pliocene-Lower Pleistocene; Roda, 1965), having a base-failure under the foot of the slope and an estimated thickness of landslide body of at least 80-100 m. Bluish-grey clays change rapidly in stratigraphic continuity into yellow sand deposits, whose extension and thickness (no more than 30 m) are more developed toward the northern and eastern edges of the landslide. Both constitute regressive sediments forming the Niscemi Mesa (dating back to Middle Pleistocene; Roda, 1965). This is locally shifted by faults and persistent landslide activity (Grasso & La Manna, 1993). Separate aquifers are generally found at the claysand contacts. The recent landslide has destabilised, with noticeable geometrical coincidence, an ancient landslide body, whose only mobilisation which was well documented (the recent urbanisation of Niscemi dates back to the XVI century) occurred on March 18, 1790 (Nava, 1792), involving the same district (SS. Croci). The old mobilisation was similarly slow, although it lasted longer (the paroxystic stage developed over two days), had a higher velocity (presumably having a maximum of about 4-6 m/h) and higher vertical displacement (about 20 m along the main scarp and a similar amount of uplift along the foot, displacing a part of the valley bottom plain). It also had a base failure without toe overlap, whose morphological effects are still evident. The old mobilisation was accompanied by precursory deep sound and small post-paroxysmal thermal phenomena (maccaluba) so impressive at the time as to be falsely ascribed to volcanic (ignivoma) phenomena (Masaracchio, 1790; Maugeri, 1869). In order to understand better the role of rainfall in recent landslide reactivation, an analysis of daily rainfall records of the past century and their cumulate values on several sets of days was carried out. The results show that, although only the data of the last century were considered, there have been several tens of higher rainfall events than those which preceded the landslide. During the same time no significant earthquakes have been reported in the site (>VIIIMCS). Bearing these first results in mind and hypothesizing that variation in inherent factors (such as pore-water pressure and strength properties) of such a large and thick landslide body cannot be changed by a short heavy rainfall, this study aimed to establish other factors inducing the landslide, such are predisposing and active geological structures.