A lumped conceptual rainfall-runoff model based on the MIKE 11/NAM modeling code has been applied to the 372 km2 Tsitika River Watershed. The model was constructed on the basis of readily accessible data of precipitation, temperature, land cover and topography, and was calibrated against the river discharge at the watershed outlet. Several validation tests were carried out and model performance evaluated in terms of the water balance error and the agreement of general hydrograph shape. Sensitivity analysis was conducted to explore the variability of simulation response produced by uncertainty associated with parameter values. In general, the model captured the dynamics of river discharge moderately well with most problems arising during the simulations of the snowmelt season. The differences between observations and model output were attributed to the insufficient spatial coverage of meteorological input data, errors in model structure resulting from a simplified model set-up, and errors caused by inevitable simplifications of temporal and spatial characteristics of the hydrologic behaviour of a very complex natural system.