Land surface models (LSMs) simulate vertical fluxes, including evapotranspiration, in a rigorous manner, and are included in atmospheric models, including Regional and Global Circulation Models (RCMs and GCMs). Large-scale hydrological models on the other hand simulate the lateral processes that generate streamflow. Coupling of the two models (referred to as a hydrological land surface model) has the potential to combine the strengths of each. The MESH model developed at Environment Canada is such model that combines the Canadian Land Surface Scheme (CLASS) with a distributed hydrological model called WATFLOOD. In this thesis, the performance of the MESH model was explored using two different runoff generation schemes (i.e., elementary and enhanced runoff generation) and with a priori parameter values and with parameter calibration. The model was tested in the White Gull creek Basin located in the boreal forest, central Saskatchewan using meteorology and flux data recorded at two monitoring stations within the basin for driving and validation. Application of the model with a priori parameter values without calibration resulted in poor performance in simulating both streamflow and evapotranspiration while optimization to calibrate the model to the observed streamflow resulted in a good performance. Streamflow simulation with enhanced runoff generation included performed even better. The optimal model configuration was taken forward for a detailed parameter sensitivity analysis. Univariate analysis was used for pre-screening the parameter space to eliminate insensitive parameters, and subsequently multivariate analysis was performed for a subset of parameters. Vegetation parameters were more identifiable when an objective function measuring the fit to observed latent heat flux was used than when measuring the fit to streamflow. Physiographic and topographic parameters were more identifiable when a streamflow objective function was used. Streamflow was more sensitive to parameter variability than latent heat flux. The use of multiple objective functions to simultaneously constrain the model was explored. Selection of objective function had no significant effect on the simulated evapotranspiration but had some influence on streamflow. Using NSE objective function with streamflow was found to be the most effective way of identifying the best model runs. The additional constraints imposed by evapotranspiration had no impact on the results.