The Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis (TMPA) provides three-hourly, 0.25°x0.25° gridded precipitation estimates based on intercalibrated passive microwave estimates, microwave-calibrated infrared estimates, and monthly raingauge analyses for the 10 years 1998-2007 over the latitude band 50°N-50°S. The uniform, fine-scale coverage is appealing to a wide range of users, but we continue to work to summarize the dataset in useful ways and compare these results to surface-based data to validate the dataset as possible. One critical application for TMPA data is to characterize the occurrence of extreme precipitation events across the globe, realizing that such a short record cannot yield truly climatological statistics. Multi-scale analyses of precipitation intensity percentiles as a function of duration and coverage are reviewed to identify typical patterns. Seasonal and interannual variations also will be reviewed. As expected, interannual variations are closely tied to ENSO events. For regions and scales where it is feasible, surface-based data are similarly processed and analyzed to validate the TMPA data, most extensively in the continental U.S. We address the well-known difficulty in assessing useful agreement between fine-scale precipitation data sets by applying fuzzy verification metrics. Specifically, we examine accumulations, unconditional histograms, and conditional histograms accumulated over matched regions across a range of space and time scales. This work reiterates the difficulty (shared by all such fine-scale estimators) in capturing small-scale features, and gives quantitative statements about the improvement for larger scales. Besides providing guidance for current and prospective users, these results are helping shape the development of error estimators for such precipitation algorithms.