We present results of mercury (Hg) in surface waters and soils and an analysis of satellite imagery from the Tapajós River basin, Brazilian Amazon, and the Reserva Garimpeira do Tapajós, the legal gold mining district of the basin. Hg bound to suspended sediment was roughly 600 and 200 times the concentration of dissolved Hg per litre of water, in impacted and pristine areas, respectively. Suspended sediments thus represent the major pathway of river-borne Hg. Median concentrations of Hg in suspended load from both impacted and pristine waters were 134 ppb, and 80% of samples were below 300ppb-in the range of naturally occurring surficial materials in the tropics. Regionally, riverine Hg fluxes were proportional to the concentration of total suspended solids. This shows that the dominant source of Hg is the sediment itself rather than anthropogenic mercury discharge from the small-scale mines. To independently test this conclusion, a mass balance was performed. A conservative calculation of the annual export of mercury (Hg) from the Creporí River (a minimum) was 1.6 tonnes for the year 1998-it could be significantly larger. This amount of Hg is difficult to account for by anthropogenic discharge alone, confirming that enhanced physical erosion caused by sluicing and dredging operations is the dominant source of Hg. We therefore conclude that gold mining operations are primarily responsible for elevated Hg concentrations. The dominant source of contamination is not, however, the loss of Hg in the gold amalgamation process. Rather, the disturbance and mobilization of large quantities of Hg-rich sediment and floodplain soil into the water column during mining operations is the source of contamination. These findings shift the focus of remediation and prevention efforts away from Hg control toward soil and sediment erosion control. The minimization or elimination of Hg losses in the mining process remains important for the health of local peoples and environments, but keeping basin soils and sediments in place would be a much more effective means of minimizing Hg fluxes to the region's rivers. To gain a spatial and historical perspective on the source and extent of emissions, satellite imagery was used. We were able to reconstruct historical mining activity, locate impacted areas, and estimate historical Hg fluxes with the imagery. To do so, the spectral characteristics of satellite images were calibrated to the concentration of suspended sediment in the rivers, which, in turn, is proportional to the Hg concentration. This analysis shows that mining-induced sediment plumes have been a dominant source of sediment to the Tapajós River system for decades. As well, the intensity and location of these emissions has varied through time. For example, sediment discharge from the Creporí River was greater in 1985 than in 1998; and the tributaries on the west bank of the Tapajós were actively being mined in 1985 but had been abandoned in 1998. This type of information should greatly assist in understanding original and ongoing sources of emissions, and in managing prevention and remediation efforts.