The applicability of satellite precipitation products like the TRMM multi-satellite precipitation analysis (TMPA) products has been limited in its applicability to real-time hydrologic prediction (e.g., of floods and droughts) by the need for gauge-based adjustments. However, recent advances in the TMPA algorithms and data streams now offer encouragement regarding their potential use for real-time hydrologic prediction without gauge adjustment. This potential is of particular interest in regions (e.g. Africa) where in situ gauge networks are sparse. We evaluate how changes in the TMPA precipitation retrieval algorithms have affected the products in the context of flood and drought prediction, and quantify the uncertainty of streamflow predictions resulting from TMPA real-time (TMPA-RT) errors with respect to the gauge corrected product. We conduct hydrologic simulations using the Variable Infiltration Capacity (VIC) model over 12 global river basins within the TMPA domain for the 2002 to 2010 period. For evaluation purposes, the model is forced by the TMPA-V6 research product (gauge adjusted) and TMPA-RT (no gauge adjustment). Other meteorological forcings are taken from an independent dataset developed at Princeton University. The daily model output is routed to the main river outlet for each basin using a one quarter degree routing network. The model soil parameters are calibrated against observed discharge over these basins. Results based on TMPA-V6 are used as a benchmark for both examining the evolution of the real-time product algorithms and quantifying the uncertainty of hydrologic prediction based on the latest algorithm.