Natural concentrations of 238U and ??234U values were determined in estuarine surface waters and pore waters of the Amazon and Fly (Papua New Guinea) Rivers to investigate U transport phenomena across river-dominated land-sea margins. Discharge from large, tropical rivers is a major source of dissolved and solid materials transported to the oceans, and are important in defining not only oceanic mass budgets, but also terrestrial weathering rates. On the Amazon shelf, salinity-property plots of dissolved organic carbon, pH and total suspended matter revealed two vastly contrasting water masses that were energetically mixed. In this mixing zone, the distribution of uranium was highly non-conservative and exhibited extensive removal from the water column. Uranium removal was most pronounced within a salinity range of 0-16.6, and likely the result of scavenging and flocculation reactions with inorganic (i.e., Fe/Mn oxides) and organic colloids/particles. Removal of uranium may also be closely coupled to exchange and resuspension processes at the sediment/water interface. An inner-shelf pore water profile indicated the following diagenetic processes: extensive (???1 m) zones of Fe(III) - and, to a lesser degree, Mn(IV) - reduction in the absence of significant S(II) concentrations appeared to facilitate the formation of various authigenic minerals (e.g., siderite, rhodocrosite and uraninite). The pore water dissolved 238U profile co-varied closely with Mn(II). Isotopic variations as evidenced in ??234U pore waters values from this site revealed information on the origin and history of particulate uranium. Only after a depth of about 1 m did the ??234U value approach unity (secular equilibrium), denoting a residual lattice bound uranium complex that is likely an upper-drainage basin weathering product. This suggests that the enriched ??234U values represent a riverine surface complexation product that is actively involved in Mn-Fe diagenetic cycles and surface complexation reactions. In the Fly River estuary, 238U appears to exhibit a reasonably conservative distribution as a function of salinity. The absence of observed U removal does not necessarily imply non-reactivity, but instead may record an integration of concurrent U removal and release processes. There is not a linear correlation between ??234U vs. 1/ 238U that would imply simple two component mixing. It is likely that resuspension of bottom sediments, prolonged residence times in the lower reaches of the Fly River, and energetic particle-colloid interactions contribute to the observed estuarine U distribution. The supply of uranium discharged from humid, tropical river systems to the sea appears to be foremost influenced by particle/water interactions that are ultimately governed by the particular physiographic and hydrologic characteristics of an estuary. ?? 2004 Elsevier Ltd. All rights reserved.