We constructed a whole carbon budget for a catchment in the Western Amazon Basin, combining drainage water analyses with eddy covariance (EC) measured terrestrial CO2 fluxes. As fluvial C export can represent permanent C export it must be included in assessments of whole site C balance, but it is rarely done. The footprint area of the flux tower is drained by two small streams (~5–7 km2) from which we measured the dissolved inorganic carbon (DIC), dissolved organic carbon (DOC), particulate organic carbon (POC) export, and CO2 efflux. The EC measurements showed the site C balance to be +0.7 ± 9.7 Mg C ha−1 yr−1 (a source to the atmosphere) and fluvial export was 0.3 ± 0.04 Mg C ha−1 yr−1. Of the total fluvial loss 34% was DIC, 37% DOC, and 29% POC. The wet season was most important for fluvial C export. There was a large uncertainty associated with the EC results and with previous biomass plot studies (−0.5 ± 4.1 Mg C ha−1 yr−1); hence, it cannot be concluded with certainty whether the site is C sink or source. The fluvial export corresponds to only 3–7% of the uncertainty related to the site C balance; thus, other factors need to be considered to reduce the uncertainty and refine the estimated C balance. However, stream C export is significant, especially for almost neutral sites where fluvial loss may determine the direction of the site C balance. The fate of C downstream then dictates the overall climate impact of fluvial export.