Digital maps of geology, ground slope, and dormant landslides are combined statistically in a geographic information system (GIS) to identify sites of future landsliding over a broad area. The resulting index number, a continuous variable, predicts a range of susceptibility both within and between existing landslides. Spatial resolution of the index can be as fine as that of the slope map, and areal coverage is limited only by the extent of the input data. Susceptibility is defined for each geologic-map unit as the spatial frequency of the unit occupied by dormant landslides, adjusted locally by ground slope. Susceptibility of terrain between landslides is calculated for each one-degree slope interval as the percentage of grid cells that coincide with the failures. Susceptibility within landslides is the same percentage times the comparative frequency of recent failures within and outside the old landslides. We tested the model in an 872 km 2 urban area in California, using 120 geologic units, a 30-m digital elevation model, 6714 dormant landslide deposits, 1192 recent landslides, and ARC/INFO software. The method could generate a similar map for any area where the necessary digital-map data are available.