The chemical reactivity of uranium was investigated across estuarine gradients from two of the world's largest river systems: the Amazon and Mississippi. Concentrations of dissolved (<0.45 ??m) uranium (U) were measured in surface waters of the Amazon shelf during rising (March 1990), flood (June 1990) and low (November 1991) discharge regimes. The dissolved U content was also examined in surface waters collected across estuarine gradients of the Mississippi outflow region during April 1992, August 1993, and November (1993). All water samples were analyzed for U by isotope dilution inductively coupled plasma mass spectrometry (ICP-MS). In Amazon shelf surface waters uranium increased nonconservatively from about 0.01 ??g l-1 at the river's mouth to over 3 ??g l-1 at the distal site, irrespective of river discharge stage. Observed large-scale U removal at salinities generally less than 15 implies a) that riverine dissolved U was extensively adsorbed by freshly-precipitated hydrous metal oxides (e.g., FeOOH, MnO2) as a result of flocculation and aggregation, and b) that energetic resuspension and reworking of shelf sediments and fluid muds on the Amazon shelf released a chemically reactive particle/colloid to the water column which can further scavenge dissolved U across much of the estuarine gradient. In contrast, the estuarine chemistry of U is inconclusive within surface waters of the Mississippi shelf-break region. U behavior is most likely controlled less by traditional sorption and/or desorption reactions involving metal oxides or colloids than by the river's variable discharge regime (e.g., water parcel residence time during estuarine mixing, nature of particulates, sediment storage and resuspension in the confined lower river), and plume dispersal. Mixing of the thin freshwater lens into ambient seawater is largely defined by wind-driven rather than physical processes. As a consequence, in the Mississippi outflow region uranium predominantly displays ConserVative behavior; removal is evident only during anomalous river discharge regimes. 'Products-approach' mixing experiments conducted during the Flood of 1993 suggest the importance of small particles and/or colloids in defining a depleted U versus salinity distribution.