Streamflow was calculated from meteorological data using the UBC Watershed Model for 12 heterogeneous watersheds in British Columbia. These watersheds varied in drainage area, climate, topography, soil type vegetation and geology, which indicates that statistical regionalization of streamflow is likely to be unreliable. The model results indicated that an average set of model parameters could be applied to all the watersheds, provided that representative meteorological data was available and the impermeable fraction of the watersheds could be determined from independent sources such as natural vegetative cover and surficial geology. The averaged model parameter set gave Nash-Sutcliffe statistical results only slightly lower than results from a full calibration. Therefore the method appears to be useful for complete pre-calibration of ungauged watersheds so that flows can be generated for these ungauged sites. The method has been successfully tested on an independent watershed in British Columbia and another watershed in the Himalayas.