River discharge is a key variable to quantify the water cycle, its fluxes and stocks at different scales, from local scale for the efficient management of water resources to global scale for the monitoring of climate change. However, gathering reliable, long term and consistent information on river discharges worldwide or on large transboundary river basins is an extremely complex task, if ever achievable, as Hydrologic Services in different countries have heterogeneous acquisition strategies and data policies. Therefore, developing Earth Observation (EO) techniques for the measurement or estimation of river discharge is a major challenge. A key question deals with the possibility of deriving river discharge values from EO surface variables (width, level, slope, velocity, the only one accessible through EO) without any in situ measurement. Based on a literature study and original developments, the possibilities of estimating water surface variables using remote-sensing techniques have been explored, mainly radar altimetry as well as across-track and along-track interferometry. Then a method has been developed, based in a first phase on the equations of the uniform regime, in order to estimate river discharge from these surface variables only. The river section is simplified assuming to have a rectangular cross-section represented by its mean bottom level and width. Another hypothesis is made on a constant coefficient linking water surface velocity to the river section mean velocity. Based on a set of surface variables measurement at different dates and hydrological regimes, the methods estimates the values of the mean bottom level and mean Manning coefficient. Therefore, to be applied, the method requires a reasonable number of measurements along the complete hydrological cycle. The method has been developed and tested on a dataset of measurements realized on several stations on the Amazon basin (HyBAM ANA-IRD Project). Surface velocities and surface width are provided through ADCP measurements while water level and longitudinal river slopes are provided by in situ monitoring of levelled gauging stations and relevant technique to derive the longitudinal profile and slope. This method has been tested on different stations of the Amazon basin and gives satisfactory results on some of them but discrepancies on others. It appears that time varying surface slope on Amazon stations is in contradiction with the uniform hypothesis; therefore the method has been adapted to a non uniform flow configuration. The new method give relevant results on simulated data and further development are on-going to increase robustness of this method to noisy data.