Hydrological models are widely applied to make informed decisions in water resources planning and management. However, the representation of hydrological processes in a model is subject to uncertainty due to the conceptualization of those processes. The choice of models is a trade-off among availability of data and understanding of processes in the catchment. In this study, we compared the performance of two hydrological models, the conceptual GR4JSG and the process based J2000 hydrological model in a glaciated alpine catchment in the Himalayan region. Both models were adapted to simulate the glacier melt runoff in which seasonal snowmelt and glacier icemelt occurs. Regarding the spatial heterogeneity, the catchments are represented by numbers of hydrological response units (or functional units in GR4JSG). Both models were calibrated for the period 1986-1991 and validated for the period 1992-1997. The models were able to represent the overall hydrograph with a Nash Sutcliffe Efficiency (NSE) in the J2000 model of 0.84 and 0.87 for calibration and validation, respectively, and for the GR4JSG model, the NSE was 0.87 and 0.89 for the calibration and validate periods. Although the models simulated baseflow and medium range flows (recession and rising limbs) reasonably well, the peak flows were underestimated in some instances. On average, the percentage bias is below 3% in both models. The hydrograph suggested that the GR4JSG tends to underestimate observed hydrograph during pre-monsoon season. Similarly, GR4JSG tends to overestimate July-September compared to J2000. The glacier melt contribution to stream flow is about 13% and 17% for GR4JSG and J2000 respectively. The variation can be expected due to conceptualization of hydrological processes in both models.