Three datasets of long-term, hourly precipitation with high spatial resolution were developed for Southern California to support detailed hydrologic modeling of flash flood occurrence. The precipitation estimates based on three methods of varying complexity are compared in this presentation: (a) an interpolation of hourly observations onto a 1km2 grid using the monthly climatological spatial distribution through PRISM; (b) a simplified numerical model of topographically-driven precipitation with 3x3 km2 spatial resolution; (b) precipitation estimates from a mesoscale numerical model (RSM) with 10x10km2 spatial resolution; The estimates spanned the historical period from October 1948 through April 2005. The three methods are compared in terms of spatial distribution and magnitude of wet-season and monthly average precipitation, of regional and small watershed (50 to 400 km2) seasonal climatology, and frequency of hourly precipitation at different precipitation rates. The comparison indicates similarities in the spatial distribution, monthly climatology, and sub-region characteristics among Southern California mountain ranges, although the magnitude varies among models. The combination of these different methods forms a multi-model ensemble of long-term, high resolution precipitation estimates appropriate for further analysis and hydrologic studies.