The profitability of hydropower in Costa Rica is affected by soil erosion and sedimentation in dam reservoirs, which are in turn influenced by land use, infiltration and aquifer interactions with surface water. In order to foster the provision and payment of Hydrological Environmental Services (HES), a quantitative assessment of the impact of specific land uses on the functioning of drainage-basins is required. The present paper aims to study the water balance partitioning in a volcanic coffee agroforestry micro-basin (1 km2, steep slopes) in Costa Rica, as a first step towards evaluating sediment or contaminant loads. The main hydrological processes were monitored during one year, using flume, eddy-covariance flux tower, soil water profiles and piezometers. A new Hydro-SVAT lumped model is proposed, that balances SVAT (Soil Vegetation Atmosphere Transfer) and basin-reservoir routines. The purpose of such a coupling was to achieve a trade-off between the expected performance of ecophysiological and hydrological models, which are often employed separately and at different spatial scales, either the plot or the basin. The calibration of the model to perform streamflow yielded a NS coefficient equal to 0.80, while the validation of the water balance partitioning was consistent with the independent measurements of actual evapotranspiration (R2=0.79, energy balance closed independently), soil water content (R2=0.49) and water table level (R2=0.90). An uncertainty analysis showed that the streamflow modelling was precise for nearly every time step, while a sensitivity analysis revealed which parameters mostly affected model precision, depending on the season. It was observed that 64% of the incident rainfall R flowed out of the basin as streamflow, 25% as evapotranspiration and the remaining 11% was attributed to deep percolation. The model indicated an interception loss equal to 4% of R, a surface runoff of 5% and an infiltration component of 91%. The modelled streamflow