In order to study the influence of potential evapotranspiration on daily flow simulation results for hydrological models in the northeast alpine region, the widely applied SWAT model was chosen as an example, through comparative study on Ougen River Basin, using Penman-Monteith model, Priestley-Taylor model, Hargreaves model, Shuttleworth-Wallace model and modified 20cm evaporating dish observed data to compute the potential evapotranspiration (PET), then the influence of SWAT models on the simulation results was studied by different PET drivers. Research results showed that adopting subbasin scale for SWAT model to PET interpolation was quiet rough, and the calculated results cannot reflect the spatial distribution of PET specifically, the average temperature of alpine forest areas was less than 17.8℃ from January to February and also in December every year, including March sometimes. In different subbasins covered by large forest, the change of monthly average PET was little with the variation of forest species types and distribution areas, although differences of terrain on SWAT model were considered, it didnot show a high daily flow precision. Penman-Monteith model without consideration of terrain pattern, and Hargreaves model and Shuttleworth-Wallace model considered terrain pattern had good simulation results, while Penman-Monteith model fitted the best with Ens values of 0.651 and 0.686 in calibration and validation periods, respectively, which showed that the Penman-Monteith model suited better in alpine forest areas to calculate the potential evapotranspiration.