The choice of data resolution in rainfall-runoff modelling primarily depends on its availability. However, if high-resolution data is available, it is important to choose a model with appropriate complexity since a simpler model may be insufficient to extract additional information available in high-resolution data. On the other hand, when applying a complex model to low-resolution data, some of the model components may become redundant, and computational effort and time may not be judiciously utilized. To assess the impact of temporal resolution and model complexity on model performance, a lumped model structure with variable complexity was developed and tested on two mountainous catchments using rainfall and streamflow data available at a 15-min resolution. Downscaled resolutions of hourly, daily, and weekly time steps and altered model structures were tested to observe the impact of reduced resolution and reduced complexity on model performance. The general observation is that if high resolution data is available, modelling at that resolution and downscaling the output to a desired lower resolution preserves the accuracy better than running the model at the lower resolution. Comparison of results from the two test catchments showed that a slow responding catchment can be modelled at a lower resolution and by a simpler model. However, for the catchment that produces rapidly rising and receding hydrographs with sharper peaks, a complex model may be warranted, and the data should also be of finer resolution containing sufficient information to support such a model.