Water quality models are increasingly used to predict nutrient losses from anthropogenically impacted catchments, but comprehensive model calibration and transfer of model parameters are often difficult. The objectives of this study are (i) to assess the Hydrological Predictions for the Environment (HYPE) model for simulating runoff and inorganic nitrogen (IN) leaching in nested and spatially heterogeneous mesoscale catchments in central Germany and (ii) to investigate the temporal and spatial variations of IN leaching. A multi-site and multi-objective calibration approach using the parameter estimation tool PEST was employed. Results showed that parameters related to evapotranspiration were most sensitive for runoff simulation. The nitrogen balance was controlled by plant uptake and denitrification. Runoff was well reproduced for both calibration (1994–1999) and validation (1999–2004) periods at all three gauging stations, with the lowest Nash–Sutcliffe efficiency (NSE) of 0·86 and bias (PBIAS) of less than 15%. In addition, the model was tested successfully to predict the dynamics of soil water storage during extreme climatological events. IN concentrations showed increase in magnitude and decrease in dynamics from upstream to downstream of the catchment, reflecting the combined effects of nutrient inputs and IN in-stream retention. The dynamics of monthly IN loads were well represented by the model during the whole simulation period, with the lowest NSE and PBIAS of 0·69 and 33%, respectively. Results revealed seasonal dynamics of IN leaching due to the combined effects of hydrological and biogeochemical processes, which is consistent with runoff dynamics. This indicates the controlling role of hydrology on nitrogen transport processes. Copyright © 2013 John Wiley & Sons, Ltd.