HBV-N is a conceptual process-based model for simulation of transformation and transport of nitrogen on the catchment scale. This paper presents further development with focus on the response function and calibration procedures. Evaluation of the model routines was made in 12 test basins in southern Sweden (without or with only few lakes). Previous versions of HBV-N included a HBV version with a single reservoir in the response function. The presented results show that both nitrogen concentrations and water discharge simulations improved when a second reservoir was introduced. The two-reservoir model was found to be more well-posed for description of residence-times and flow paths. On average, this resulted in an increase in explained variance (R-2) for nitrogen concentrations by 0.3. Multiple-response split-sample calibration was found to further improve the model performance and reliability. In previous applications HBV-N has been applied by using single-response calibration. However, simultaneous calibration of water discharge and nitrogen improved the R2 for nitrogen concentrations by about 0.1 (range 0.02-0.25), but did not affect the simulation of water discharge. This new calibration strategy forces the hydrological parameters to a new optimum, and reduces the level of uncertainty for both hydrochemical and hydrological modelling.