Forecasts of future landsliding currently rely on "ground truth" inventories to validate and refine quantitative or qualitative predictive models. Acceptance of inventory maps as deterministic truth, however, neglects errors and uncertainties inherent in the inventory process. A well-documented evaluation of landslide inventory maps of a 300 km2 area in northern Italy prepared by three groups of geomorphologists revealed that landslide polygon positional mismatch between maps was in the 55-65% range, whereas the positional mismatch was 80% for superposition of all three maps. Additional discrepancies must exist with landslide inventories that also include classification, volume, activity and other characteristics. Simply understanding the general nature of the uncertainty and variability associated with inventory maps is a substantial challenge; however, the basis for quantifying the uncertainty and variability requires independent reassessment of all aspects of the inventory process or acceptance of some theoretical a priori value for variance. Evaluation of some hazards (e.g., surface fault rupture) is done with apparent consensus of location and substantial attention devoted to the uncertainty associated with estimates of activity (e.g., slip rate). The scale of inventory mapping is substantially greater for landslides than for active faults because of the number and variety of landslides, the variable degree of activity within an individual landslide, and the fact that landslides are secondary hazards triggered by primary processes. Possible improvements might include 1) combining the maps by several independent inventory teams to produce empirical probabilistic maps, 2) using widely applicable statistical distributions of landslide size to estimate the numbers of landslides likely to have been missed in an inventory, and 3) using physicsbased watershed scale models of landslide occurrence to identify areas that may have been overlooked in inventories or which may be susceptible to sliding under future rare or unprecedented conditions.