There is paleohydrographic evidence that current empirical models of Probable Maximum Precipitation (PMP) may significantly overestimate actual PMP at higher elevations in Colorado. As a result, engineering standards adopted for structures such as dams may be overly stringent and unnecessarily costly. In an exploratory effort to develop alternate and more accurate models of PMP in this region, a prototypical methodology based on mesoscale ensemble modeling has been developed for estimating extreme precipitation in Colorado's complex terrain. In this approach, the Regional Atmospheric Modeling System (RAMS) is applied to numerous extreme precipitation events that have occurred in Colorado in the last 40+ years. For each case, interactive grid nesting is used to simulate the synoptic, mesoscale and cloud-resolving scales, respectively, over the western U.S., the Colorado region, and the meso-beta-scale area of the extreme precipitation event. Large scale analysis of observational data are used for initialization and time-dependent boundary conditions. The ensemble simulations for each case consist of various estimates of initial soil moisture fields and/or atmospheric perturbations such as enhanced precipitable water. An alternate ensemble simulation approach is utilized wherein the initial meteorological fields are shifted relative to the topography, allowing the specific region of forcing in a given case to be shifted to other regions in Colorado. From many such ensemble simulations based on historical cases, a large number of plausible extreme storm scenarios are simulated over Colorado's complex terrain. From the spatial and elevational dependence of extreme precipitation contained in these simulated outcomes, a prototypical method for objectively inferring PMP at any location in the Colorado Rocky Mountains is described.