Accurate gridded precipitation data is of major importance in hydrologic modelling. In areaswith complex topography and a sparse network of precipitation stations, it can be difficult toobtain gridded data of sufficient quality. As a result, precipitation data is biased. If notcorrected, these biases translate into the results of hydrologic modelling. In this work, thesebiases were calculated on a watershed basis within the Fraser River Basin, BC. For that,runoff of 38 watersheds was simulated using a simple water balance model and thencompared to observed runoff at gauge stations, referring to the same watershedboundaries. Observed streamflow data is decently accurate and therefore suited to calculatecorrection factors. With this method, it was possible to account for the regional patterns andvariations within the complex Fraser River Basin.Since the estimation of evapotranspiration is a major uncertainty of this approach, runoffwas computed with four different potential evapotranspiration (PET) methods. The modeloriginally runs with the Hamon method. Additionally, the Thornthwaite, Jensen-Haise andHargreaves methods were implemented. The result is a range of bias. Further, theconfidence of the direction of bias (positive or negative bias/ over- or underestimation ofprecipitation) was calculated. Most parts of the basin suffer from an underestimation ofprecipitation data. In the rain shadow areas of the basin, precipitation is too high. Themethods mostly agree on under- and overestimation. In conclusion, the biggest influencingfactor on bias is weather a watershed is located on the windward or the leeward side of aMountain Range, indicating different meteorological processes.Mean biases based on the four methods were calculated for every watershed on an annualbasis. The biases range from under predictions of -57% to over predictions of 99%, resultingin correction factors between 0.5 and 5.3.