Greenhouse gas fluxes (CO 2 , CH 4 , and N 2 O) from African streams and rivers are under-represented in global datasets, resulting in uncertainties in their contributions to regional and global budgets. We conducted yearlong sampling of 59 sites in a nested-catchment design in the Mara River, Kenya in which fluxes were quantified and their underlying controls assessed. We estimated annual basin-scale greenhouse gas emissions from measured in-stream gas concentrations, modeled gas transfer velocities, and determined the sensitivity of upscaling to discharge. Based on the total annual CO 2 -equivalent emissions calculated from global warming potentials (GWP), the Mara basin was a net greenhouse gas source (294 AE 35 Gg CO 2 eq yr À1 ). Lower-order streams (1-3) contributed 81% of the total fluxes, and higher stream orders (4-8) contributed 19%. Cropland-draining streams also exhibited higher fluxes compared to forested streams. Seasonality in stream discharge affected stream widths (and stream area) and gas exchange rates, strongly influencing the basin-wide annual flux, which was 10 times higher during the high and medium discharge periods than the low discharge period. The basinwide estimate was underestimated by up to 36% if discharge was ignored, and up to 37% for lower stream orders. Future research should therefore include seasonality in stream surface areas in upscaling procedures to better constrain basin-wide fluxes. Given that agricultural activities are a major factor increasing riverine greenhouse gas fluxes in the study region, increased conversion of forests and agricultural intensification has the possibility of increasing the contribution of the African continent to global greenhouse gas sources.