Landslides are a widespread, frequent, and costly hazard in Seattle and the Puget Sound Region of Washington. Four deaths and property damage amounting to tens-of-millions of dollars resulted from hundreds of landslides that occurred during the winters of 1995-96 and 1996-97. Rainfall and snowmelt triggered many shallow earth slides as well as several large rotational and translational slides. Many slides transformed to debris flows and caused significant property damage downslope from their source areas. To aid the City of Seattle in planning emergency response and land use, we are developing GIS-based methods to map the probability of landslide initiation and runout for storms similar to those that triggered slides in 1996-97. We use field studies and physically based models to assess "relative" slope stability throughout the area and then analyze the correlation between historic landslide occurrence and relative slope stability to estimate the probability of landslide initiation for storms like those in 1996-97 throughout the area. We also use field studies to determine the uncertainty in the extent of runout and then use a probabilistic model to combine this information with the topography and initiation probabilities to determine probabilities of runout throughout the area. For a test area in Seattle, we correlated the degree of slope stability with locations of recent landslide sources, which occupied 4.2% of steep-slope (>10°) areas on the coastal bluff. For example, our analysis indicated that 55% of the cells on steep slopes had greater than 1% chance of failure and that 8.8% had greater than 5% chance of failure, based on uniform pore-pressure conditions, representative strength parameters, and a simple model for colluvium thickness. Additional work on colluvium thickness and bluff hydrology is aimed at improving the delineation of failure-prone slopes. Runout angles (along path, from top of source to terminus of deposit) of 264 recent slides and flows on coastal bluffs between Seattle and Everett ranged from 3° to 57° with a mean of 27.5° and a standard deviation of 9.3°, independent of source materials and consistent with studies elsewhere. The resulting map of probabilities successfully predicted the downslope extent of landslide runout along most sections of the bluff.