Snow- and glacier melt are important contributors to river discharge in high-elevated areas of the Himalayan region. Thus, it is important that the key processes controlling snow and glacier accumulation and melting, are well represented in hydrological models. In this study, the sensitivity of modelled discharge to different snowmelt parameterizations was evaluated. A distributed hydrological model that operated on a 1 × 1 km2 grid at a daily time resolution was applied to a high-elevated mountainous basin, the Upper Beas basin in Indian Himalaya, including several sub-basins with a varying degree of glacier covered areas. The snowmelt was calculated using (i) a temperature-index method, (ii) an enhanced temperature-index method including a shortwave radiation term, and (iii) an energy balance method. All model configurations showed similar performance at daily, seasonal, and annual timescales and a lower performance for the validation period than for the calibration period; a main reason being the failure to capture the observed negative trend in annual discharge in the validation period. The results suggest that model performance is more sensitive to the precipitation input, i.e. interpolation method than to the choice of snowmelt routine. The paper highlights the challenges related to the lack of high quality data sets in mountainous regions, which are those areas globally with most water resources.